{ "editionDate": "2026-10-19", "generatedAt": "2026-10-19T07:00:00-04:00", "dataCutoff": "2026-10-18T23:59:59-04:00", "historicalDelta": { "version": "1.0", "currentEdition": "2026-10-19", "previousEdition": "2026-10-12", "generatedAt": "2026-10-19T07:00:00-04:00", "trackedCategories": [ "models", "availability", "prices", "benchmarks", "freeTiers" ], "materialChangeCount": 6, "counts": { "added": 2, "changed": 4, "removed": 0 }, "categories": { "models": { "id": "models", "label": "Models", "description": "New, removed or materially revised model records.", "status": "changed", "changeCount": 4, "previousTracked": 36, "currentTracked": 37, "unchangedCount": 33, "changes": [ { "category": "models", "type": "changed", "key": "Alibaba|Qwen-Image-2.1-Turbo", "label": "Alibaba Qwen-Image-2.1-Turbo", "summary": "Alibaba Qwen-Image-2.1-Turbo: released changed", "fields": [ { "field": "released", "previous": "2026-10-10", "current": "2026-10-15", "delta": null, "direction": "changed" } ], "previous": { "origin": "China", "provider": "Alibaba", "model": "Qwen-Image-2.1-Turbo", "released": "2026-10-10", "availability": "Open Weights (Hugging Face / ModelScope), Alibaba Cloud API", "context": "8,192 text & reference tokens", "inputPrice": 0.014, "cachedInputPrice": 0.007, "outputPrice": 0.014, "priceCurrency": "USD", "strength": "Accelerated 8-step Flow Matching Euler diffusion with 64-channel RGBA VAE for native 2K transparent image synthesis (418 ms on H100)", "modalities": "Text-to-Image, Multi-Reference Editing, Transparent RGBA", "caveat": "Commercial self-hosting requires enterprise commercial license agreement from Alibaba Tongyi Lab.", "sourceId": "qwen-image-2-1-turbo-announcement" }, "current": { "origin": "China", "provider": "Alibaba", "model": "Qwen-Image-2.1-Turbo", "released": "2026-10-15", "availability": "Open Weights (Hugging Face / ModelScope), Alibaba Cloud API", "context": "8,192 text & reference tokens", "inputPrice": 0.014, "cachedInputPrice": 0.007, "outputPrice": 0.014, "priceCurrency": "USD", "strength": "Accelerated 8-step Flow Matching Euler diffusion with 64-channel RGBA VAE for native 2K transparent image synthesis (418 ms on H100)", "modalities": "Text-to-Image, Multi-Reference Editing, Transparent RGBA", "caveat": "Commercial self-hosting requires enterprise commercial license agreement from Alibaba Tongyi Lab.", "sourceId": "openai-gpt6-ga-whitepaper" }, "sourceId": "openai-gpt6-ga-whitepaper" }, { "category": "models", "type": "changed", "key": "Anthropic|Claude 3.8 (Workspaces & Motion)", "label": "Anthropic Claude 3.8 (Workspaces & Motion)", "summary": "Anthropic Claude 3.8 (Workspaces & Motion): released, modalities, and strength changed", "fields": [ { "field": "released", "previous": "2026-10-11", "current": "2026-10-16", "delta": null, "direction": "changed" }, { "field": "modalities", "previous": "Text, Vision, Bash Sandbox, Terminal Tools, Motion SVG/WebGL", "current": "Text, Vision, Code, Filesystem Actions, WebGL Canvas", "delta": null, "direction": "changed" }, { "field": "strength", "previous": "Native POSIX bash sandboxing, collaborative workspace artifacts GA, and Motion dynamic animated WebGL visualization engine", "current": "Native host POSIX microVM sandboxing, collaborative workspace artifacts GA, and Motion dynamic animated WebGL visualization engine", "delta": null, "direction": "changed" } ], "previous": { "origin": "US", "provider": "Anthropic", "model": "Claude 3.8 (Workspaces & Motion)", "released": "2026-10-11", "availability": "Anthropic API, Claude Enterprise, Claude Pro", "context": "200,000 native", "inputPrice": 3, "cachedInputPrice": 0.75, "outputPrice": 15, "priceCurrency": "USD", "strength": "Native POSIX bash sandboxing, collaborative workspace artifacts GA, and Motion dynamic animated WebGL visualization engine", "modalities": "Text, Vision, Bash Sandbox, Terminal Tools, Motion SVG/WebGL", "caveat": "Autonomous script execution restricted by seccomp-bpf filter policy; multi-step actions require loopback sandboxing.", "sourceId": "anthropic-claude-motion-usage-policy" }, "current": { "origin": "US", "provider": "Anthropic", "model": "Claude 3.8 (Workspaces & Motion)", "released": "2026-10-16", "availability": "Anthropic API, Claude Enterprise, Claude Pro", "context": "200,000 native", "inputPrice": 3, "cachedInputPrice": 0.75, "outputPrice": 15, "priceCurrency": "USD", "strength": "Native host POSIX microVM sandboxing, collaborative workspace artifacts GA, and Motion dynamic animated WebGL visualization engine", "modalities": "Text, Vision, Code, Filesystem Actions, WebGL Canvas", "caveat": "Higher token price tier; requires hardware virtualization (KVM) support on agent host.", "sourceId": "anthropic-claude-motion-posix-policy" }, "sourceId": "anthropic-claude-motion-posix-policy" }, { "category": "models", "type": "added", "key": "Google DeepMind|Gemini 3.8 (Flash & Pro)", "label": "Google DeepMind Gemini 3.8 (Flash & Pro)", "summary": "Google DeepMind Gemini 3.8 (Flash & Pro) added to tracking", "fields": [], "previous": null, "current": { "origin": "US", "provider": "Google DeepMind", "model": "Gemini 3.8 (Flash & Pro)", "released": "2026-10-15", "availability": "Google AI Studio, Vertex AI, Gemini Advanced", "context": "2,000,000 native", "inputPrice": 0.075, "cachedInputPrice": 0.019, "outputPrice": 0.3, "priceCurrency": "USD", "strength": "Native 2M context with sub-125ms live multimodal streaming, top-ranked general reasoning (65.4 on General365), and integrated SynthID Universal provenance watermarks", "modalities": "Text, Vision, Audio, Video, Code, Tool Use", "caveat": "Full 2M token context requires enterprise API tier; dynamic thinking mode tokens billed at standard output rates.", "sourceId": "google-gemini-38-multimodal-release" }, "sourceId": "google-gemini-38-multimodal-release" }, { "category": "models", "type": "changed", "key": "OpenAI|GPT-6 (Sol & Luna)", "label": "OpenAI GPT-6 (Sol & Luna)", "summary": "OpenAI GPT-6 (Sol & Luna): released and strength changed", "fields": [ { "field": "released", "previous": "2026-10-11", "current": "2026-10-14", "delta": null, "direction": "changed" }, { "field": "strength", "previous": "Interactive Intelligent UI canvas generation with declarative component trees and dual-engine reasoning (Sol 180 tok/s, Luna full-modality)", "current": "Dual-engine speculative decoding runtime (Sol 180 tok/s, Luna deep reasoning verifier) with native declarative Intelligent UI component trees", "delta": null, "direction": "changed" } ], "previous": { "origin": "US", "provider": "OpenAI", "model": "GPT-6 (Sol & Luna)", "released": "2026-10-11", "availability": "ChatGPT Free, Plus, Pro, Enterprise, OpenAI API", "context": "256,000 native", "inputPrice": 0.35, "cachedInputPrice": 0.08, "outputPrice": 1.4, "priceCurrency": "USD", "strength": "Interactive Intelligent UI canvas generation with declarative component trees and dual-engine reasoning (Sol 180 tok/s, Luna full-modality)", "modalities": "Text, Vision, Audio, UI Components, Canvas Tools", "caveat": "Mandatory C2PA metadata injection enforced on all visual generation; synthetic election asset restrictions active.", "sourceId": "openai-chatgpt-intelligent-ui-gpt6" }, "current": { "origin": "US", "provider": "OpenAI", "model": "GPT-6 (Sol & Luna)", "released": "2026-10-14", "availability": "ChatGPT Enterprise, Pro, Plus, API", "context": "256,000 native", "inputPrice": 0.35, "cachedInputPrice": 0.08, "outputPrice": 1.4, "priceCurrency": "USD", "strength": "Dual-engine speculative decoding runtime (Sol 180 tok/s, Luna deep reasoning verifier) with native declarative Intelligent UI component trees", "modalities": "Text, Vision, Audio, UI Components, Canvas Tools", "caveat": "Mandatory C2PA metadata injection enforced on all visual generation; synthetic election asset restrictions active.", "sourceId": "openai-gpt6-ga-whitepaper" }, "sourceId": "openai-gpt6-ga-whitepaper" } ] }, "availability": { "id": "availability", "label": "Availability", "description": "Endpoint, region, alias, access and lifecycle changes.", "status": "changed", "changeCount": 1, "previousTracked": 36, "currentTracked": 37, "unchangedCount": 35, "changes": [ { "category": "availability", "type": "changed", "key": "OpenAI|GPT-6 (Sol & Luna)", "label": "OpenAI GPT-6 (Sol & Luna)", "summary": "OpenAI GPT-6 (Sol & Luna): availability changed", "fields": [ { "field": "availability", "previous": "ChatGPT Free, Plus, Pro, Enterprise, OpenAI API", "current": "ChatGPT Enterprise, Pro, Plus, API", "delta": null, "direction": "changed" } ], "previous": { "origin": "US", "provider": "OpenAI", "model": "GPT-6 (Sol & Luna)", "released": "2026-10-11", "availability": "ChatGPT Free, Plus, Pro, Enterprise, OpenAI API", "context": "256,000 native", "inputPrice": 0.35, "cachedInputPrice": 0.08, "outputPrice": 1.4, "priceCurrency": "USD", "strength": "Interactive Intelligent UI canvas generation with declarative component trees and dual-engine reasoning (Sol 180 tok/s, Luna full-modality)", "modalities": "Text, Vision, Audio, UI Components, Canvas Tools", "caveat": "Mandatory C2PA metadata injection enforced on all visual generation; synthetic election asset restrictions active.", "sourceId": "openai-chatgpt-intelligent-ui-gpt6" }, "current": { "origin": "US", "provider": "OpenAI", "model": "GPT-6 (Sol & Luna)", "released": "2026-10-14", "availability": "ChatGPT Enterprise, Pro, Plus, API", "context": "256,000 native", "inputPrice": 0.35, "cachedInputPrice": 0.08, "outputPrice": 1.4, "priceCurrency": "USD", "strength": "Dual-engine speculative decoding runtime (Sol 180 tok/s, Luna deep reasoning verifier) with native declarative Intelligent UI component trees", "modalities": "Text, Vision, Audio, UI Components, Canvas Tools", "caveat": "Mandatory C2PA metadata injection enforced on all visual generation; synthetic election asset restrictions active.", "sourceId": "openai-gpt6-ga-whitepaper" }, "sourceId": "openai-gpt6-ga-whitepaper" } ] }, "prices": { "id": "prices", "label": "Prices", "description": "API token prices, paid-plan terms and published promotions.", "status": "changed", "changeCount": 1, "previousTracked": 47, "currentTracked": 48, "unchangedCount": 47, "changes": [ { "category": "prices", "type": "added", "key": "api|Google DeepMind|Gemini 3.8 (Flash & Pro)", "label": "Google DeepMind Gemini 3.8 (Flash & Pro)", "summary": "Google DeepMind Gemini 3.8 (Flash & Pro) price or plan terms added to tracking", "fields": [], "previous": null, "current": { "inputPrice": 0.075, "outputPrice": 0.3, "cachedInputPrice": 0.019, "priceCurrency": "USD" }, "sourceId": "google-gemini-38-multimodal-release" } ] }, "benchmarks": { "id": "benchmarks", "label": "Benchmarks", "description": "Comparable score, rank, coverage or methodology-status changes.", "status": "unchanged", "changeCount": 0, "previousTracked": 57, "currentTracked": 57, "unchangedCount": 57, "changes": [] }, "freeTiers": { "id": "freeTiers", "label": "Free tiers", "description": "Published free-plan availability, limits and eligibility terms.", "status": "unchanged", "changeCount": 0, "previousTracked": 2, "currentTracked": 2, "unchangedCount": 2, "changes": [] } }, "timeline": [ { "category": "models", "type": "changed", "key": "Alibaba|Qwen-Image-2.1-Turbo", "label": "Alibaba Qwen-Image-2.1-Turbo", "summary": "Alibaba Qwen-Image-2.1-Turbo: released changed", "fields": [ { "field": "released", "previous": "2026-10-10", "current": "2026-10-15", "delta": null, "direction": "changed" } ], "previous": { "origin": "China", "provider": "Alibaba", "model": "Qwen-Image-2.1-Turbo", "released": "2026-10-10", "availability": "Open Weights (Hugging Face / ModelScope), Alibaba Cloud API", "context": "8,192 text & reference tokens", "inputPrice": 0.014, "cachedInputPrice": 0.007, "outputPrice": 0.014, "priceCurrency": "USD", "strength": "Accelerated 8-step Flow Matching Euler diffusion with 64-channel RGBA VAE for native 2K transparent image synthesis (418 ms on H100)", "modalities": "Text-to-Image, Multi-Reference Editing, Transparent RGBA", "caveat": "Commercial self-hosting requires enterprise commercial license agreement from Alibaba Tongyi Lab.", "sourceId": "qwen-image-2-1-turbo-announcement" }, "current": { "origin": "China", "provider": "Alibaba", "model": "Qwen-Image-2.1-Turbo", "released": "2026-10-15", "availability": "Open Weights (Hugging Face / ModelScope), Alibaba Cloud API", "context": "8,192 text & reference tokens", "inputPrice": 0.014, "cachedInputPrice": 0.007, "outputPrice": 0.014, "priceCurrency": "USD", "strength": "Accelerated 8-step Flow Matching Euler diffusion with 64-channel RGBA VAE for native 2K transparent image synthesis (418 ms on H100)", "modalities": "Text-to-Image, Multi-Reference Editing, Transparent RGBA", "caveat": "Commercial self-hosting requires enterprise commercial license agreement from Alibaba Tongyi Lab.", "sourceId": "openai-gpt6-ga-whitepaper" }, "sourceId": "openai-gpt6-ga-whitepaper" }, { "category": "models", "type": "changed", "key": "Anthropic|Claude 3.8 (Workspaces & Motion)", "label": "Anthropic Claude 3.8 (Workspaces & Motion)", "summary": "Anthropic Claude 3.8 (Workspaces & Motion): released, modalities, and strength changed", "fields": [ { "field": "released", "previous": "2026-10-11", "current": "2026-10-16", "delta": null, "direction": "changed" }, { "field": "modalities", "previous": "Text, Vision, Bash Sandbox, Terminal Tools, Motion SVG/WebGL", "current": "Text, Vision, Code, Filesystem Actions, WebGL Canvas", "delta": null, "direction": "changed" }, { "field": "strength", "previous": "Native POSIX bash sandboxing, collaborative workspace artifacts GA, and Motion dynamic animated WebGL visualization engine", "current": "Native host POSIX microVM sandboxing, collaborative workspace artifacts GA, and Motion dynamic animated WebGL visualization engine", "delta": null, "direction": "changed" } ], "previous": { "origin": "US", "provider": "Anthropic", "model": "Claude 3.8 (Workspaces & Motion)", "released": "2026-10-11", "availability": "Anthropic API, Claude Enterprise, Claude Pro", "context": "200,000 native", "inputPrice": 3, "cachedInputPrice": 0.75, "outputPrice": 15, "priceCurrency": "USD", "strength": "Native POSIX bash sandboxing, collaborative workspace artifacts GA, and Motion dynamic animated WebGL visualization engine", "modalities": "Text, Vision, Bash Sandbox, Terminal Tools, Motion SVG/WebGL", "caveat": "Autonomous script execution restricted by seccomp-bpf filter policy; multi-step actions require loopback sandboxing.", "sourceId": "anthropic-claude-motion-usage-policy" }, "current": { "origin": "US", "provider": "Anthropic", "model": "Claude 3.8 (Workspaces & Motion)", "released": "2026-10-16", "availability": "Anthropic API, Claude Enterprise, Claude Pro", "context": "200,000 native", "inputPrice": 3, "cachedInputPrice": 0.75, "outputPrice": 15, "priceCurrency": "USD", "strength": "Native host POSIX microVM sandboxing, collaborative workspace artifacts GA, and Motion dynamic animated WebGL visualization engine", "modalities": "Text, Vision, Code, Filesystem Actions, WebGL Canvas", "caveat": "Higher token price tier; requires hardware virtualization (KVM) support on agent host.", "sourceId": "anthropic-claude-motion-posix-policy" }, "sourceId": "anthropic-claude-motion-posix-policy" }, { "category": "models", "type": "added", "key": "Google DeepMind|Gemini 3.8 (Flash & Pro)", "label": "Google DeepMind Gemini 3.8 (Flash & Pro)", "summary": "Google DeepMind Gemini 3.8 (Flash & Pro) added to tracking", "fields": [], "previous": null, "current": { "origin": "US", "provider": "Google DeepMind", "model": "Gemini 3.8 (Flash & Pro)", "released": "2026-10-15", "availability": "Google AI Studio, Vertex AI, Gemini Advanced", "context": "2,000,000 native", "inputPrice": 0.075, "cachedInputPrice": 0.019, "outputPrice": 0.3, "priceCurrency": "USD", "strength": "Native 2M context with sub-125ms live multimodal streaming, top-ranked general reasoning (65.4 on General365), and integrated SynthID Universal provenance watermarks", "modalities": "Text, Vision, Audio, Video, Code, Tool Use", "caveat": "Full 2M token context requires enterprise API tier; dynamic thinking mode tokens billed at standard output rates.", "sourceId": "google-gemini-38-multimodal-release" }, "sourceId": "google-gemini-38-multimodal-release" }, { "category": "models", "type": "changed", "key": "OpenAI|GPT-6 (Sol & Luna)", "label": "OpenAI GPT-6 (Sol & Luna)", "summary": "OpenAI GPT-6 (Sol & Luna): released and strength changed", "fields": [ { "field": "released", "previous": "2026-10-11", "current": "2026-10-14", "delta": null, "direction": "changed" }, { "field": "strength", "previous": "Interactive Intelligent UI canvas generation with declarative component trees and dual-engine reasoning (Sol 180 tok/s, Luna full-modality)", "current": "Dual-engine speculative decoding runtime (Sol 180 tok/s, Luna deep reasoning verifier) with native declarative Intelligent UI component trees", "delta": null, "direction": "changed" } ], "previous": { "origin": "US", "provider": "OpenAI", "model": "GPT-6 (Sol & Luna)", "released": "2026-10-11", "availability": "ChatGPT Free, Plus, Pro, Enterprise, OpenAI API", "context": "256,000 native", "inputPrice": 0.35, "cachedInputPrice": 0.08, "outputPrice": 1.4, "priceCurrency": "USD", "strength": "Interactive Intelligent UI canvas generation with declarative component trees and dual-engine reasoning (Sol 180 tok/s, Luna full-modality)", "modalities": "Text, Vision, Audio, UI Components, Canvas Tools", "caveat": "Mandatory C2PA metadata injection enforced on all visual generation; synthetic election asset restrictions active.", "sourceId": "openai-chatgpt-intelligent-ui-gpt6" }, "current": { "origin": "US", "provider": "OpenAI", "model": "GPT-6 (Sol & Luna)", "released": "2026-10-14", "availability": "ChatGPT Enterprise, Pro, Plus, API", "context": "256,000 native", "inputPrice": 0.35, "cachedInputPrice": 0.08, "outputPrice": 1.4, "priceCurrency": "USD", "strength": "Dual-engine speculative decoding runtime (Sol 180 tok/s, Luna deep reasoning verifier) with native declarative Intelligent UI component trees", "modalities": "Text, Vision, Audio, UI Components, Canvas Tools", "caveat": "Mandatory C2PA metadata injection enforced on all visual generation; synthetic election asset restrictions active.", "sourceId": "openai-gpt6-ga-whitepaper" }, "sourceId": "openai-gpt6-ga-whitepaper" }, { "category": "availability", "type": "changed", "key": "OpenAI|GPT-6 (Sol & Luna)", "label": "OpenAI GPT-6 (Sol & Luna)", "summary": "OpenAI GPT-6 (Sol & Luna): availability changed", "fields": [ { "field": "availability", "previous": "ChatGPT Free, Plus, Pro, Enterprise, OpenAI API", "current": "ChatGPT Enterprise, Pro, Plus, API", "delta": null, "direction": "changed" } ], "previous": { "origin": "US", "provider": "OpenAI", "model": "GPT-6 (Sol & Luna)", "released": "2026-10-11", "availability": "ChatGPT Free, Plus, Pro, Enterprise, OpenAI API", "context": "256,000 native", "inputPrice": 0.35, "cachedInputPrice": 0.08, "outputPrice": 1.4, "priceCurrency": "USD", "strength": "Interactive Intelligent UI canvas generation with declarative component trees and dual-engine reasoning (Sol 180 tok/s, Luna full-modality)", "modalities": "Text, Vision, Audio, UI Components, Canvas Tools", "caveat": "Mandatory C2PA metadata injection enforced on all visual generation; synthetic election asset restrictions active.", "sourceId": "openai-chatgpt-intelligent-ui-gpt6" }, "current": { "origin": "US", "provider": "OpenAI", "model": "GPT-6 (Sol & Luna)", "released": "2026-10-14", "availability": "ChatGPT Enterprise, Pro, Plus, API", "context": "256,000 native", "inputPrice": 0.35, "cachedInputPrice": 0.08, "outputPrice": 1.4, "priceCurrency": "USD", "strength": "Dual-engine speculative decoding runtime (Sol 180 tok/s, Luna deep reasoning verifier) with native declarative Intelligent UI component trees", "modalities": "Text, Vision, Audio, UI Components, Canvas Tools", "caveat": "Mandatory C2PA metadata injection enforced on all visual generation; synthetic election asset restrictions active.", "sourceId": "openai-gpt6-ga-whitepaper" }, "sourceId": "openai-gpt6-ga-whitepaper" }, { "category": "prices", "type": "added", "key": "api|Google DeepMind|Gemini 3.8 (Flash & Pro)", "label": "Google DeepMind Gemini 3.8 (Flash & Pro)", "summary": "Google DeepMind Gemini 3.8 (Flash & Pro) price or plan terms added to tracking", "fields": [], "previous": null, "current": { "inputPrice": 0.075, "outputPrice": 0.3, "cachedInputPrice": 0.019, "priceCurrency": "USD" }, "sourceId": "google-gemini-38-multimodal-release" } ], "unchangedCategories": [ "benchmarks", "freeTiers" ], "methodology": { "comparisonType": "Field-level day-over-day comparison", "sourceOnlyChangesIgnored": true, "modelFields": [ "released", "context", "modalities", "strength", "origin" ], "availabilityFields": [ "availability" ], "apiPriceFields": [ "inputPrice", "outputPrice", "cachedInputPrice", "priceCurrency" ], "benchmarkFields": [ "score", "value", "rank", "coverage", "status", "result", "unit" ], "freeTierFields": [ "price", "limit", "availability", "eligibility", "feature", "notes" ] } }, "trendTimeline": { "version": "1.0", "asOf": "2026-10-19", "timezone": "America/New_York", "availableHistoryStart": "2026-10-12", "availableHistoryEnd": "2026-10-19", "availableEditionCount": 2, "availableCalendarDays": 8, "defaultWindow": 7, "requestedWindows": [ 7, 30, 90 ], "groups": { "models": { "label": "Models and availability", "description": "Model additions, removals, capability-record revisions, endpoint access and lifecycle movement.", "categories": [ "models", "availability" ] }, "prices": { "label": "Prices and free tiers", "description": "API prices, subscriptions, promotions, free-plan availability and published quota changes.", "categories": [ "prices", "freeTiers" ] }, "benchmarks": { "label": "Benchmarks", "description": "Comparable scores, coverage, ranking and methodology-status movement.", "categories": [ "benchmarks" ] } }, "providers": [ "Alibaba", "Anthropic", "Google DeepMind", "OpenAI" ], "comparisonCount": 1, "totalEventCount": 6, "windows": { "7": { "id": "7d", "days": 7, "label": "7-day", "requestedStartDate": "2026-10-13", "endDate": "2026-10-19", "coverageStartDate": "2026-10-19", "coverageEndDate": "2026-10-19", "editionCount": 1, "coveredCalendarDays": 1, "missingEditionDays": 0, "complete": false, "coveragePercent": 14, "coverageLabel": "1 of 7 calendar days represented by 1 preserved editions", "eventCount": 6, "providers": [ "Alibaba", "Anthropic", "Google DeepMind", "OpenAI" ], "indicators": { "models": { "group": "models", "label": "Models and availability", "description": "Model additions, removals, capability-record revisions, endpoint access and lifecycle movement.", "categories": [ "models", "availability" ], "eventCount": 5, "counts": { "added": 1, "changed": 4, "removed": 0 }, "status": "expanding", "symbol": "↑", "statusLabel": "Expanding", "latestEventDate": "2026-10-19" }, "prices": { "group": "prices", "label": "Prices and free tiers", "description": "API prices, subscriptions, promotions, free-plan availability and published quota changes.", "categories": [ "prices", "freeTiers" ], "eventCount": 1, "counts": { "added": 1, "changed": 0, "removed": 0 }, "status": "expanding", "symbol": "↑", "statusLabel": "Expanding", "latestEventDate": "2026-10-19" }, "benchmarks": { "group": "benchmarks", "label": "Benchmarks", "description": "Comparable scores, coverage, ranking and methodology-status movement.", "categories": [ "benchmarks" ], "eventCount": 0, "counts": { "added": 0, "changed": 0, "removed": 0 }, "status": "stable", "symbol": "→", "statusLabel": "Stable", "latestEventDate": null } }, "dailyActivity": [ { "date": "2026-10-19", "displayDate": "Monday, October 19, 2026", "total": 6, "models": 5, "prices": 1, "benchmarks": 0 } ], "events": [ { "id": "2026-10-19|models|changed|Alibaba|Qwen-Image-2.1-Turbo", "date": "2026-10-19", "previousEdition": "2026-10-12", "group": "models", "category": "models", "provider": "Alibaba", "type": "changed", "key": "Alibaba|Qwen-Image-2.1-Turbo", "label": "Alibaba Qwen-Image-2.1-Turbo", "summary": "Alibaba Qwen-Image-2.1-Turbo: released changed", "fields": [ { "field": "released", "previous": "2026-10-10", "current": "2026-10-15", "delta": null, "direction": "changed" } ], "sourceId": "openai-gpt6-ga-whitepaper" }, { "id": "2026-10-19|models|changed|Anthropic|Claude 3.8 (Workspaces & Motion)", "date": "2026-10-19", "previousEdition": "2026-10-12", "group": "models", "category": "models", "provider": "Anthropic", "type": "changed", "key": "Anthropic|Claude 3.8 (Workspaces & Motion)", "label": "Anthropic Claude 3.8 (Workspaces & Motion)", "summary": "Anthropic Claude 3.8 (Workspaces & Motion): released, modalities, and strength changed", "fields": [ { "field": "released", "previous": "2026-10-11", "current": "2026-10-16", "delta": null, "direction": "changed" }, { "field": "modalities", "previous": "Text, Vision, Bash Sandbox, Terminal Tools, Motion SVG/WebGL", "current": "Text, Vision, Code, Filesystem Actions, WebGL Canvas", "delta": null, "direction": "changed" }, { "field": "strength", "previous": "Native POSIX bash sandboxing, collaborative workspace artifacts GA, and Motion dynamic animated WebGL visualization engine", "current": "Native host POSIX microVM sandboxing, collaborative workspace artifacts GA, and Motion dynamic animated WebGL visualization engine", "delta": null, "direction": "changed" } ], "sourceId": "anthropic-claude-motion-posix-policy" }, { "id": "2026-10-19|models|added|Google DeepMind|Gemini 3.8 (Flash & Pro)", "date": "2026-10-19", "previousEdition": "2026-10-12", "group": "models", "category": "models", "provider": "Google DeepMind", "type": "added", "key": "Google DeepMind|Gemini 3.8 (Flash & Pro)", "label": "Google DeepMind Gemini 3.8 (Flash & Pro)", "summary": "Google DeepMind Gemini 3.8 (Flash & Pro) added to tracking", "fields": [], "sourceId": "google-gemini-38-multimodal-release" }, { "id": "2026-10-19|models|changed|OpenAI|GPT-6 (Sol & Luna)", "date": "2026-10-19", "previousEdition": "2026-10-12", "group": "models", "category": "models", "provider": "OpenAI", "type": "changed", "key": "OpenAI|GPT-6 (Sol & Luna)", "label": "OpenAI GPT-6 (Sol & Luna)", "summary": "OpenAI GPT-6 (Sol & Luna): released and strength changed", "fields": [ { "field": "released", "previous": "2026-10-11", "current": "2026-10-14", "delta": null, "direction": "changed" }, { "field": "strength", "previous": "Interactive Intelligent UI canvas generation with declarative component trees and dual-engine reasoning (Sol 180 tok/s, Luna full-modality)", "current": "Dual-engine speculative decoding runtime (Sol 180 tok/s, Luna deep reasoning verifier) with native declarative Intelligent UI component trees", "delta": null, "direction": "changed" } ], "sourceId": "openai-gpt6-ga-whitepaper" }, { "id": "2026-10-19|availability|changed|OpenAI|GPT-6 (Sol & Luna)", "date": "2026-10-19", "previousEdition": "2026-10-12", "group": "models", "category": "availability", "provider": "OpenAI", "type": "changed", "key": "OpenAI|GPT-6 (Sol & Luna)", "label": "OpenAI GPT-6 (Sol & Luna)", "summary": "OpenAI GPT-6 (Sol & Luna): availability changed", "fields": [ { "field": "availability", "previous": "ChatGPT Free, Plus, Pro, Enterprise, OpenAI API", "current": "ChatGPT Enterprise, Pro, Plus, API", "delta": null, "direction": "changed" } ], "sourceId": "openai-gpt6-ga-whitepaper" }, { "id": "2026-10-19|prices|added|api|Google DeepMind|Gemini 3.8 (Flash & Pro)", "date": "2026-10-19", "previousEdition": "2026-10-12", "group": "prices", "category": "prices", "provider": "Google DeepMind", "type": "added", "key": "api|Google DeepMind|Gemini 3.8 (Flash & Pro)", "label": "Google DeepMind Gemini 3.8 (Flash & Pro)", "summary": "Google DeepMind Gemini 3.8 (Flash & Pro) price or plan terms added to tracking", "fields": [], "sourceId": "google-gemini-38-multimodal-release" } ] }, "30": { "id": "30d", "days": 30, "label": "30-day", "requestedStartDate": "2026-09-20", "endDate": "2026-10-19", "coverageStartDate": "2026-10-12", "coverageEndDate": "2026-10-19", "editionCount": 2, "coveredCalendarDays": 8, "missingEditionDays": 6, "complete": false, "coveragePercent": 27, "coverageLabel": "8 of 30 calendar days represented by 2 preserved editions", "eventCount": 6, "providers": [ "Alibaba", "Anthropic", "Google DeepMind", "OpenAI" ], "indicators": { "models": { "group": "models", "label": "Models and availability", "description": "Model additions, removals, capability-record revisions, endpoint access and lifecycle movement.", "categories": [ "models", "availability" ], "eventCount": 5, "counts": { "added": 1, "changed": 4, "removed": 0 }, "status": "expanding", "symbol": "↑", "statusLabel": "Expanding", "latestEventDate": "2026-10-19" }, "prices": { "group": "prices", "label": "Prices and free tiers", "description": "API prices, subscriptions, promotions, free-plan availability and published quota changes.", "categories": [ "prices", "freeTiers" ], "eventCount": 1, "counts": { "added": 1, "changed": 0, "removed": 0 }, "status": "expanding", "symbol": "↑", "statusLabel": "Expanding", "latestEventDate": "2026-10-19" }, "benchmarks": { "group": "benchmarks", "label": "Benchmarks", "description": "Comparable scores, coverage, ranking and methodology-status movement.", "categories": [ "benchmarks" ], "eventCount": 0, "counts": { "added": 0, "changed": 0, "removed": 0 }, "status": "stable", "symbol": "→", "statusLabel": "Stable", "latestEventDate": null } }, "dailyActivity": [ { "date": "2026-10-12", "displayDate": "Monday, October 12, 2026", "total": 0, "models": 0, "prices": 0, "benchmarks": 0 }, { "date": "2026-10-19", "displayDate": "Monday, October 19, 2026", "total": 6, "models": 5, "prices": 1, "benchmarks": 0 } ], "events": [ { "id": "2026-10-19|models|changed|Alibaba|Qwen-Image-2.1-Turbo", "date": "2026-10-19", "previousEdition": "2026-10-12", "group": "models", "category": "models", "provider": "Alibaba", "type": "changed", "key": "Alibaba|Qwen-Image-2.1-Turbo", "label": "Alibaba Qwen-Image-2.1-Turbo", "summary": "Alibaba Qwen-Image-2.1-Turbo: released changed", "fields": [ { "field": "released", "previous": "2026-10-10", "current": "2026-10-15", "delta": null, "direction": "changed" } ], "sourceId": "openai-gpt6-ga-whitepaper" }, { "id": "2026-10-19|models|changed|Anthropic|Claude 3.8 (Workspaces & Motion)", "date": "2026-10-19", "previousEdition": "2026-10-12", "group": "models", "category": "models", "provider": "Anthropic", "type": "changed", "key": "Anthropic|Claude 3.8 (Workspaces & Motion)", "label": "Anthropic Claude 3.8 (Workspaces & Motion)", "summary": "Anthropic Claude 3.8 (Workspaces & Motion): released, modalities, and strength changed", "fields": [ { "field": "released", "previous": "2026-10-11", "current": "2026-10-16", "delta": null, "direction": "changed" }, { "field": "modalities", "previous": "Text, Vision, Bash Sandbox, Terminal Tools, Motion SVG/WebGL", "current": "Text, Vision, Code, Filesystem Actions, WebGL Canvas", "delta": null, "direction": "changed" }, { "field": "strength", "previous": "Native POSIX bash sandboxing, collaborative workspace artifacts GA, and Motion dynamic animated WebGL visualization engine", "current": "Native host POSIX microVM sandboxing, collaborative workspace artifacts GA, and Motion dynamic animated WebGL visualization engine", "delta": null, "direction": "changed" } ], "sourceId": "anthropic-claude-motion-posix-policy" }, { "id": "2026-10-19|models|added|Google DeepMind|Gemini 3.8 (Flash & Pro)", "date": "2026-10-19", "previousEdition": "2026-10-12", "group": "models", "category": "models", "provider": "Google DeepMind", "type": "added", "key": "Google DeepMind|Gemini 3.8 (Flash & Pro)", "label": "Google DeepMind Gemini 3.8 (Flash & Pro)", "summary": "Google DeepMind Gemini 3.8 (Flash & Pro) added to tracking", "fields": [], "sourceId": "google-gemini-38-multimodal-release" }, { "id": "2026-10-19|models|changed|OpenAI|GPT-6 (Sol & Luna)", "date": "2026-10-19", "previousEdition": "2026-10-12", "group": "models", "category": "models", "provider": "OpenAI", "type": "changed", "key": "OpenAI|GPT-6 (Sol & Luna)", "label": "OpenAI GPT-6 (Sol & Luna)", "summary": "OpenAI GPT-6 (Sol & Luna): released and strength changed", "fields": [ { "field": "released", "previous": "2026-10-11", "current": "2026-10-14", "delta": null, "direction": "changed" }, { "field": "strength", "previous": "Interactive Intelligent UI canvas generation with declarative component trees and dual-engine reasoning (Sol 180 tok/s, Luna full-modality)", "current": "Dual-engine speculative decoding runtime (Sol 180 tok/s, Luna deep reasoning verifier) with native declarative Intelligent UI component trees", "delta": null, "direction": "changed" } ], "sourceId": "openai-gpt6-ga-whitepaper" }, { "id": "2026-10-19|availability|changed|OpenAI|GPT-6 (Sol & Luna)", "date": "2026-10-19", "previousEdition": "2026-10-12", "group": "models", "category": "availability", "provider": "OpenAI", "type": "changed", "key": "OpenAI|GPT-6 (Sol & Luna)", "label": "OpenAI GPT-6 (Sol & Luna)", "summary": "OpenAI GPT-6 (Sol & Luna): availability changed", "fields": [ { "field": "availability", "previous": "ChatGPT Free, Plus, Pro, Enterprise, OpenAI API", "current": "ChatGPT Enterprise, Pro, Plus, API", "delta": null, "direction": "changed" } ], "sourceId": "openai-gpt6-ga-whitepaper" }, { "id": "2026-10-19|prices|added|api|Google DeepMind|Gemini 3.8 (Flash & Pro)", "date": "2026-10-19", "previousEdition": "2026-10-12", "group": "prices", "category": "prices", "provider": "Google DeepMind", "type": "added", "key": "api|Google DeepMind|Gemini 3.8 (Flash & Pro)", "label": "Google DeepMind Gemini 3.8 (Flash & Pro)", "summary": "Google DeepMind Gemini 3.8 (Flash & Pro) price or plan terms added to tracking", "fields": [], "sourceId": "google-gemini-38-multimodal-release" } ] }, "90": { "id": "90d", "days": 90, "label": "90-day", "requestedStartDate": "2026-07-22", "endDate": "2026-10-19", "coverageStartDate": "2026-10-12", "coverageEndDate": "2026-10-19", "editionCount": 2, "coveredCalendarDays": 8, "missingEditionDays": 6, "complete": false, "coveragePercent": 9, "coverageLabel": "8 of 90 calendar days represented by 2 preserved editions", "eventCount": 6, "providers": [ "Alibaba", "Anthropic", "Google DeepMind", "OpenAI" ], "indicators": { "models": { "group": "models", "label": "Models and availability", "description": "Model additions, removals, capability-record revisions, endpoint access and lifecycle movement.", "categories": [ "models", "availability" ], "eventCount": 5, "counts": { "added": 1, "changed": 4, "removed": 0 }, "status": "expanding", "symbol": "↑", "statusLabel": "Expanding", "latestEventDate": "2026-10-19" }, "prices": { "group": "prices", "label": "Prices and free tiers", "description": "API prices, subscriptions, promotions, free-plan availability and published quota changes.", "categories": [ "prices", "freeTiers" ], "eventCount": 1, "counts": { "added": 1, "changed": 0, "removed": 0 }, "status": "expanding", "symbol": "↑", "statusLabel": "Expanding", "latestEventDate": "2026-10-19" }, "benchmarks": { "group": "benchmarks", "label": "Benchmarks", "description": "Comparable scores, coverage, ranking and methodology-status movement.", "categories": [ "benchmarks" ], "eventCount": 0, "counts": { "added": 0, "changed": 0, "removed": 0 }, "status": "stable", "symbol": "→", "statusLabel": "Stable", "latestEventDate": null } }, "dailyActivity": [ { "date": "2026-10-12", "displayDate": "Monday, October 12, 2026", "total": 0, "models": 0, "prices": 0, "benchmarks": 0 }, { "date": "2026-10-19", "displayDate": "Monday, October 19, 2026", "total": 6, "models": 5, "prices": 1, "benchmarks": 0 } ], "events": [ { "id": "2026-10-19|models|changed|Alibaba|Qwen-Image-2.1-Turbo", "date": "2026-10-19", "previousEdition": "2026-10-12", "group": "models", "category": "models", "provider": "Alibaba", "type": "changed", "key": "Alibaba|Qwen-Image-2.1-Turbo", "label": "Alibaba Qwen-Image-2.1-Turbo", "summary": "Alibaba Qwen-Image-2.1-Turbo: released changed", "fields": [ { "field": "released", "previous": "2026-10-10", "current": "2026-10-15", "delta": null, "direction": "changed" } ], "sourceId": "openai-gpt6-ga-whitepaper" }, { "id": "2026-10-19|models|changed|Anthropic|Claude 3.8 (Workspaces & Motion)", "date": "2026-10-19", "previousEdition": "2026-10-12", "group": "models", "category": "models", "provider": "Anthropic", "type": "changed", "key": "Anthropic|Claude 3.8 (Workspaces & Motion)", "label": "Anthropic Claude 3.8 (Workspaces & Motion)", "summary": "Anthropic Claude 3.8 (Workspaces & Motion): released, modalities, and strength changed", "fields": [ { "field": "released", "previous": "2026-10-11", "current": "2026-10-16", "delta": null, "direction": "changed" }, { "field": "modalities", "previous": "Text, Vision, Bash Sandbox, Terminal Tools, Motion SVG/WebGL", "current": "Text, Vision, Code, Filesystem Actions, WebGL Canvas", "delta": null, "direction": "changed" }, { "field": "strength", "previous": "Native POSIX bash sandboxing, collaborative workspace artifacts GA, and Motion dynamic animated WebGL visualization engine", "current": "Native host POSIX microVM sandboxing, collaborative workspace artifacts GA, and Motion dynamic animated WebGL visualization engine", "delta": null, "direction": "changed" } ], "sourceId": "anthropic-claude-motion-posix-policy" }, { "id": "2026-10-19|models|added|Google DeepMind|Gemini 3.8 (Flash & Pro)", "date": "2026-10-19", "previousEdition": "2026-10-12", "group": "models", "category": "models", "provider": "Google DeepMind", "type": "added", "key": "Google DeepMind|Gemini 3.8 (Flash & Pro)", "label": "Google DeepMind Gemini 3.8 (Flash & Pro)", "summary": "Google DeepMind Gemini 3.8 (Flash & Pro) added to tracking", "fields": [], "sourceId": "google-gemini-38-multimodal-release" }, { "id": "2026-10-19|models|changed|OpenAI|GPT-6 (Sol & Luna)", "date": "2026-10-19", "previousEdition": "2026-10-12", "group": "models", "category": "models", "provider": "OpenAI", "type": "changed", "key": "OpenAI|GPT-6 (Sol & Luna)", "label": "OpenAI GPT-6 (Sol & Luna)", "summary": "OpenAI GPT-6 (Sol & Luna): released and strength changed", "fields": [ { "field": "released", "previous": "2026-10-11", "current": "2026-10-14", "delta": null, "direction": "changed" }, { "field": "strength", "previous": "Interactive Intelligent UI canvas generation with declarative component trees and dual-engine reasoning (Sol 180 tok/s, Luna full-modality)", "current": "Dual-engine speculative decoding runtime (Sol 180 tok/s, Luna deep reasoning verifier) with native declarative Intelligent UI component trees", "delta": null, "direction": "changed" } ], "sourceId": "openai-gpt6-ga-whitepaper" }, { "id": "2026-10-19|availability|changed|OpenAI|GPT-6 (Sol & Luna)", "date": "2026-10-19", "previousEdition": "2026-10-12", "group": "models", "category": "availability", "provider": "OpenAI", "type": "changed", "key": "OpenAI|GPT-6 (Sol & Luna)", "label": "OpenAI GPT-6 (Sol & Luna)", "summary": "OpenAI GPT-6 (Sol & Luna): availability changed", "fields": [ { "field": "availability", "previous": "ChatGPT Free, Plus, Pro, Enterprise, OpenAI API", "current": "ChatGPT Enterprise, Pro, Plus, API", "delta": null, "direction": "changed" } ], "sourceId": "openai-gpt6-ga-whitepaper" }, { "id": "2026-10-19|prices|added|api|Google DeepMind|Gemini 3.8 (Flash & Pro)", "date": "2026-10-19", "previousEdition": "2026-10-12", "group": "prices", "category": "prices", "provider": "Google DeepMind", "type": "added", "key": "api|Google DeepMind|Gemini 3.8 (Flash & Pro)", "label": "Google DeepMind Gemini 3.8 (Flash & Pro)", "summary": "Google DeepMind Gemini 3.8 (Flash & Pro) price or plan terms added to tracking", "fields": [], "sourceId": "google-gemini-38-multimodal-release" } ] } }, "limitations": [ "A window is marked incomplete until preserved daily editions cover the requested calendar span without gaps.", "Activity indicators summarize verified change events; they are not model-quality rankings or investment signals.", "Pricing is never averaged across currencies, regions or incompatible billing units.", "Benchmark movement is included only when the underlying historical comparison identifies a comparable score, coverage, rank, result or methodology-status change." ] }, "pricingHistory": { "version": "1.0", "asOf": "2026-10-19", "generatedAt": "2026-10-19T07:00:00-04:00", "coverage": { "availableFrom": "2026-10-12", "availableTo": "2026-10-19", "editionCount": 2, "calendarDays": 8 }, "counts": { "total": 42, "api": 30, "plan": 8, "promotion": 2, "freeTier": 2 }, "providers": [ "Alibaba", "Alibaba Cloud", "Anthropic", "Baidu", "DeepSeek", "Google", "Google DeepMind", "Meta", "Microsoft", "Mistral AI", "Moonshot AI", "OpenAI", "Perplexity", "StepFun", "xAI" ], "currencies": [ "CNY", "Not separately published", "USD" ], "regions": [ "Not separately published" ], "sourceCounts": { "primary-current": 19, "stale": 23 }, "records": [ { "key": "api|alibaba|qwen-image-2-1-turbo|usd|not-separately-published|standard-published-tier", "type": "api", "provider": "Alibaba", "product": "Qwen-Image-2.1-Turbo", "currentTerms": { "inputPrice": 0.014, "outputPrice": 0.014, "cachedInputPrice": 0.007, "currency": "USD", "unit": "Per 1M tokens", "region": "Not separately published", "tier": "Standard published tier" }, "effectiveFrom": null, "firstObserved": "2026-10-12", "lastObserved": "2026-10-19", "lastChanged": "2026-10-12", "observationCount": 2, "periodCount": 1, "periods": [ { "firstObserved": "2026-10-12", "lastObserved": "2026-10-19", "effectiveFrom": null, "terms": { "inputPrice": 0.014, "outputPrice": 0.014, "cachedInputPrice": 0.007, "currency": "USD", "unit": "Per 1M tokens", "region": "Not separately published", "tier": "Standard published tier" }, "sourceId": "openai-gpt6-ga-whitepaper", "source": { "id": "openai-gpt6-ga-whitepaper", "title": "OpenAI GPT-6 (Sol & Luna) Dual-Engine Production Deployment Specification", "publisher": "OpenAI", "url": "https://openai.com/index/intelligent-ui", "published": "2026-10-14", "retrieved": "2026-10-19T07:00:00-04:00", "type": "Primary frontier model platform and dual-engine architecture release" } } ], "sourceId": "openai-gpt6-ga-whitepaper", "sourceValidity": { "status": "primary-current", "label": "Primary source current", "primary": true, "urlValid": true, "retrievedAt": "2026-10-19T07:00:00-04:00", "ageDays": 0, "authority": "Primary frontier model platform and dual-engine architecture release", "publisher": "OpenAI", "published": "2026-10-14" } }, { "key": "api|alibaba-cloud|qwen-2-5-coder-32b-reasoner|usd|not-separately-published|standard-published-tier", "type": "api", "provider": "Alibaba Cloud", "product": "Qwen 2.5-Coder-32B-Reasoner", "currentTerms": { "inputPrice": 0.14, "outputPrice": 0.42, "cachedInputPrice": 0.035, "currency": "USD", "unit": "Per 1M tokens", "region": "Not separately published", "tier": "Standard published tier" }, "effectiveFrom": null, "firstObserved": "2026-10-12", "lastObserved": "2026-10-19", "lastChanged": "2026-10-12", "observationCount": 2, "periodCount": 1, "periods": [ { "firstObserved": "2026-10-12", "lastObserved": "2026-10-19", "effectiveFrom": null, "terms": { "inputPrice": 0.14, "outputPrice": 0.42, "cachedInputPrice": 0.035, "currency": "USD", "unit": "Per 1M tokens", "region": "Not separately published", "tier": "Standard published tier" }, "sourceId": "alibaba-qwen-2-5-coder-reasoner-paper", "source": { "id": "alibaba-qwen-2-5-coder-reasoner-paper", "title": "Qwen 2.5-Coder-32B-Reasoner: Open-Weight Test-Time Search Code Agent with Step-Level Process Reward Supervision", "publisher": "Alibaba Cloud & Qwen Team", "url": "https://arxiv.org/abs/2609.13912", "published": "2026-09-27", "retrieved": "2026-09-28T09:00:00-04:00", "type": "Primary open-weight model technical report and repository release" } } ], "sourceId": "alibaba-qwen-2-5-coder-reasoner-paper", "sourceValidity": { "status": "primary-current", "label": "Primary source current", "primary": true, "urlValid": true, "retrievedAt": "2026-09-28T09:00:00-04:00", "ageDays": 21, "authority": "Primary open-weight model technical report and repository release", "publisher": "Alibaba Cloud & Qwen Team", "published": "2026-09-27" } }, { "key": "api|anthropic|claude-3-7-sonnet|usd|not-separately-published|standard-published-tier", "type": "api", "provider": "Anthropic", "product": "Claude 3.7 Sonnet", "currentTerms": { "inputPrice": 3, "outputPrice": 15, "cachedInputPrice": 0.3, "currency": "USD", "unit": "Per 1M tokens", "region": "Not separately published", "tier": "Standard published tier" }, "effectiveFrom": null, "firstObserved": "2026-10-12", "lastObserved": "2026-10-19", "lastChanged": "2026-10-12", "observationCount": 2, "periodCount": 1, "periods": [ { "firstObserved": "2026-10-12", "lastObserved": "2026-10-19", "effectiveFrom": null, "terms": { "inputPrice": 3, "outputPrice": 15, "cachedInputPrice": 0.3, "currency": "USD", "unit": "Per 1M tokens", "region": "Not separately published", "tier": "Standard published tier" }, "sourceId": "anthropic-claude-3-7-sonnet-system-card", "source": { "id": "anthropic-claude-3-7-sonnet-system-card", "title": "Anthropic Claude 3.7 Sonnet: Hybrid Reasoning Architecture, Extended Thinking Token Scaling, and Sub-Pixel Computer Use v2", "publisher": "Anthropic", "url": "https://www.anthropic.com/news/claude-3-7-sonnet", "published": "2026-10-05", "retrieved": "2026-10-07T09:00:00-04:00", "type": "Primary foundation model system card and technical specification" } } ], "sourceId": "anthropic-claude-3-7-sonnet-system-card", "sourceValidity": { "status": "primary-current", "label": "Primary source current", "primary": true, "urlValid": true, "retrievedAt": "2026-10-07T09:00:00-04:00", "ageDays": 12, "authority": "Primary foundation model system card and technical specification", "publisher": "Anthropic", "published": "2026-10-05" } }, { "key": "api|anthropic|claude-3-7-sonnet-hybrid-thinking-mode|usd|not-separately-published|standard-published-tier", "type": "api", "provider": "Anthropic", "product": "Claude 3.7 Sonnet (Hybrid Thinking Mode)", "currentTerms": { "inputPrice": 3, "outputPrice": 15, "cachedInputPrice": 0.3, "currency": "USD", "unit": "Per 1M tokens", "region": "Not separately published", "tier": "Standard published tier" }, "effectiveFrom": null, "firstObserved": "2026-10-12", "lastObserved": "2026-10-19", "lastChanged": "2026-10-12", "observationCount": 2, "periodCount": 1, "periods": [ { "firstObserved": "2026-10-12", "lastObserved": "2026-10-19", "effectiveFrom": null, "terms": { "inputPrice": 3, "outputPrice": 15, "cachedInputPrice": 0.3, "currency": "USD", "unit": "Per 1M tokens", "region": "Not separately published", "tier": "Standard published tier" }, "sourceId": "anthropic-claude-3-7-sonnet-system-card", "source": { "id": "anthropic-claude-3-7-sonnet-system-card", "title": "Anthropic Claude 3.7 Sonnet: Hybrid Reasoning Architecture, Extended Thinking Token Scaling, and Sub-Pixel Computer Use v2", "publisher": "Anthropic", "url": "https://www.anthropic.com/news/claude-3-7-sonnet", "published": "2026-10-05", "retrieved": "2026-10-07T09:00:00-04:00", "type": "Primary foundation model system card and technical specification" } } ], "sourceId": "anthropic-claude-3-7-sonnet-system-card", "sourceValidity": { "status": "primary-current", "label": "Primary source current", "primary": true, "urlValid": true, "retrievedAt": "2026-10-07T09:00:00-04:00", "ageDays": 12, "authority": "Primary foundation model system card and technical specification", "publisher": "Anthropic", "published": "2026-10-05" } }, { "key": "api|anthropic|claude-3-7-sonnet-standard-hybrid|usd|not-separately-published|comparison-table-tier", "type": "api", "provider": "Anthropic", "product": "Claude 3.7 Sonnet (Standard / Hybrid)", "currentTerms": { "inputPrice": null, "outputPrice": null, "cachedInputPrice": null, "currency": "USD", "unit": "Per 1M tokens", "region": "Not separately published", "tier": "Comparison table tier" }, "effectiveFrom": null, "firstObserved": "2026-10-12", "lastObserved": "2026-10-19", "lastChanged": "2026-10-12", "observationCount": 2, "periodCount": 1, "periods": [ { "firstObserved": "2026-10-12", "lastObserved": "2026-10-19", "effectiveFrom": null, "terms": { "inputPrice": null, "outputPrice": null, "cachedInputPrice": null, "currency": "USD", "unit": "Per 1M tokens", "region": "Not separately published", "tier": "Comparison table tier" }, "sourceId": "anthropic-claude-3-7-sonnet-system-card", "source": { "id": "anthropic-claude-3-7-sonnet-system-card", "title": "Anthropic Claude 3.7 Sonnet: Hybrid Reasoning Architecture, Extended Thinking Token Scaling, and Sub-Pixel Computer Use v2", "publisher": "Anthropic", "url": "https://www.anthropic.com/news/claude-3-7-sonnet", "published": "2026-10-05", "retrieved": "2026-10-07T09:00:00-04:00", "type": "Primary foundation model system card and technical specification" } } ], "sourceId": "anthropic-claude-3-7-sonnet-system-card", "sourceValidity": { "status": "primary-current", "label": "Primary source current", "primary": true, "urlValid": true, "retrievedAt": "2026-10-07T09:00:00-04:00", "ageDays": 12, "authority": "Primary foundation model system card and technical specification", "publisher": "Anthropic", "published": "2026-10-05" } }, { "key": "api|anthropic|claude-3-7-sonnet-standard-latency-mode|usd|not-separately-published|standard-published-tier", "type": "api", "provider": "Anthropic", "product": "Claude 3.7 Sonnet (Standard Latency Mode)", "currentTerms": { "inputPrice": 3, "outputPrice": 15, "cachedInputPrice": 0.3, "currency": "USD", "unit": "Per 1M tokens", "region": "Not separately published", "tier": "Standard published tier" }, "effectiveFrom": null, "firstObserved": "2026-10-12", "lastObserved": "2026-10-19", "lastChanged": "2026-10-12", "observationCount": 2, "periodCount": 1, "periods": [ { "firstObserved": "2026-10-12", "lastObserved": "2026-10-19", "effectiveFrom": null, "terms": { "inputPrice": 3, "outputPrice": 15, "cachedInputPrice": 0.3, "currency": "USD", "unit": "Per 1M tokens", "region": "Not separately published", "tier": "Standard published tier" }, "sourceId": "anthropic-claude-3-7-sonnet-system-card", "source": { "id": "anthropic-claude-3-7-sonnet-system-card", "title": "Anthropic Claude 3.7 Sonnet: Hybrid Reasoning Architecture, Extended Thinking Token Scaling, and Sub-Pixel Computer Use v2", "publisher": "Anthropic", "url": "https://www.anthropic.com/news/claude-3-7-sonnet", "published": "2026-10-05", "retrieved": "2026-10-07T09:00:00-04:00", "type": "Primary foundation model system card and technical specification" } } ], "sourceId": "anthropic-claude-3-7-sonnet-system-card", "sourceValidity": { "status": "primary-current", "label": "Primary source current", "primary": true, "urlValid": true, "retrievedAt": "2026-10-07T09:00:00-04:00", "ageDays": 12, "authority": "Primary foundation model system card and technical specification", "publisher": "Anthropic", "published": "2026-10-05" } }, { "key": "api|anthropic|claude-3-8-workspaces-motion|usd|not-separately-published|standard-published-tier", "type": "api", "provider": "Anthropic", "product": "Claude 3.8 (Workspaces & Motion)", "currentTerms": { "inputPrice": 3, "outputPrice": 15, "cachedInputPrice": 0.75, "currency": "USD", "unit": "Per 1M tokens", "region": "Not separately published", "tier": "Standard published tier" }, "effectiveFrom": null, "firstObserved": "2026-10-12", "lastObserved": "2026-10-19", "lastChanged": "2026-10-12", "observationCount": 2, "periodCount": 1, "periods": [ { "firstObserved": "2026-10-12", "lastObserved": "2026-10-19", "effectiveFrom": null, "terms": { "inputPrice": 3, "outputPrice": 15, "cachedInputPrice": 0.75, "currency": "USD", "unit": "Per 1M tokens", "region": "Not separately published", "tier": "Standard published tier" }, "sourceId": "anthropic-claude-motion-posix-policy", "source": { "id": "anthropic-claude-motion-posix-policy", "title": "Anthropic Claude Motion Engine GA and Mandatory Host POSIX MicroVM Containment Policy", "publisher": "Anthropic PBC", "url": "https://www.anthropic.com/news/motion-and-governance", "published": "2026-10-16", "retrieved": "2026-10-19T07:00:00-04:00", "type": "Primary autonomous agent execution framework and microVM containment standard" } } ], "sourceId": "anthropic-claude-motion-posix-policy", "sourceValidity": { "status": "primary-current", "label": "Primary source current", "primary": true, "urlValid": true, "retrievedAt": "2026-10-19T07:00:00-04:00", "ageDays": 0, "authority": "Primary autonomous agent execution framework and microVM containment standard", "publisher": "Anthropic PBC", "published": "2026-10-16" } }, { "key": "api|anthropic|claude-fable-5|usd|not-separately-published|standard-published-tier", "type": "api", "provider": "Anthropic", "product": "Claude Fable 5", "currentTerms": { "inputPrice": 10, "outputPrice": 50, "cachedInputPrice": null, "currency": "USD", "unit": "Per 1M tokens", "region": "Not separately published", "tier": "Standard published tier" }, "effectiveFrom": null, "firstObserved": "2026-10-12", "lastObserved": "2026-10-19", "lastChanged": "2026-10-12", "observationCount": 2, "periodCount": 1, "periods": [ { "firstObserved": "2026-10-12", "lastObserved": "2026-10-19", "effectiveFrom": null, "terms": { "inputPrice": 10, "outputPrice": 50, "cachedInputPrice": null, "currency": "USD", "unit": "Per 1M tokens", "region": "Not separately published", "tier": "Standard published tier" }, "sourceId": "anthropic-fable5", "source": { "id": "anthropic-fable5", "title": "Claude Fable 5 and Mythos 5", "publisher": "Anthropic", "url": "https://www.anthropic.com/news/claude-fable-5-mythos-5", "published": "2026-06", "retrieved": "2026-07-20T19:24:00-04:00", "type": "Primary" } } ], "sourceId": "anthropic-fable5", "sourceValidity": { "status": "stale", "label": "Source retrieval stale", "primary": true, "urlValid": true, "retrievedAt": "2026-07-20T19:24:00-04:00", "ageDays": 91, "authority": "Primary", "publisher": "Anthropic", "published": "2026-06" } }, { "key": "api|anthropic|claude-opus-5|usd|not-separately-published|standard-published-tier", "type": "api", "provider": "Anthropic", "product": "Claude Opus 5", "currentTerms": { "inputPrice": 5, "outputPrice": 25, "cachedInputPrice": null, "currency": "USD", "unit": "Per 1M tokens", "region": "Not separately published", "tier": "Standard published tier" }, "effectiveFrom": null, "firstObserved": "2026-10-12", "lastObserved": "2026-10-19", "lastChanged": "2026-10-12", "observationCount": 2, "periodCount": 1, "periods": [ { "firstObserved": "2026-10-12", "lastObserved": "2026-10-19", "effectiveFrom": null, "terms": { "inputPrice": 5, "outputPrice": 25, "cachedInputPrice": null, "currency": "USD", "unit": "Per 1M tokens", "region": "Not separately published", "tier": "Standard published tier" }, "sourceId": "anthropic-models-overview-aug9", "source": { "id": "anthropic-models-overview-aug9", "title": "Claude models overview", "publisher": "Anthropic", "url": "https://platform.claude.com/docs/en/about-claude/models/overview", "published": "Current documentation", "retrieved": "2026-08-09T22:28:00-04:00", "type": "Primary model documentation" } } ], "sourceId": "anthropic-models-overview-aug9", "sourceValidity": { "status": "stale", "label": "Source retrieval stale", "primary": true, "urlValid": true, "retrievedAt": "2026-08-09T22:28:00-04:00", "ageDays": 71, "authority": "Primary model documentation", "publisher": "Anthropic", "published": "Current documentation" } }, { "key": "api|baidu|ernie-5-0|usd|not-separately-published|standard-published-tier", "type": "api", "provider": "Baidu", "product": "ERNIE 5.0", "currentTerms": { "inputPrice": 1.4, "outputPrice": 5.6, "cachedInputPrice": null, "currency": "USD", "unit": "Per 1M tokens", "region": "Not separately published", "tier": "Standard published tier" }, "effectiveFrom": null, "firstObserved": "2026-10-12", "lastObserved": "2026-10-19", "lastChanged": "2026-10-12", "observationCount": 2, "periodCount": 1, "periods": [ { "firstObserved": "2026-10-12", "lastObserved": "2026-10-19", "effectiveFrom": null, "terms": { "inputPrice": 1.4, "outputPrice": 5.6, "cachedInputPrice": null, "currency": "USD", "unit": "Per 1M tokens", "region": "Not separately published", "tier": "Standard published tier" }, "sourceId": "baidu-ernie50", "source": { "id": "baidu-ernie50", "title": "Baidu Qianfan — ERNIE 5.0 model list and international price", "publisher": "Baidu AI Cloud", "url": "https://intl.cloud.baidu.com/en/doc/qianfan/s/7m95lyy43-intl-en", "published": "Current page", "retrieved": "2026-07-20T20:00:00-04:00", "type": "Primary" } } ], "sourceId": "baidu-ernie50", "sourceValidity": { "status": "stale", "label": "Source retrieval stale", "primary": true, "urlValid": true, "retrievedAt": "2026-07-20T20:00:00-04:00", "ageDays": 91, "authority": "Primary", "publisher": "Baidu AI Cloud", "published": "Current page" } }, { "key": "api|deepseek|deepseek-v3-5|usd|not-separately-published|standard-published-tier", "type": "api", "provider": "DeepSeek", "product": "DeepSeek-V3.5", "currentTerms": { "inputPrice": 0.14, "outputPrice": 0.28, "cachedInputPrice": 0.028, "currency": "USD", "unit": "Per 1M tokens", "region": "Not separately published", "tier": "Standard published tier" }, "effectiveFrom": null, "firstObserved": "2026-10-12", "lastObserved": "2026-10-19", "lastChanged": "2026-10-12", "observationCount": 2, "periodCount": 1, "periods": [ { "firstObserved": "2026-10-12", "lastObserved": "2026-10-19", "effectiveFrom": null, "terms": { "inputPrice": 0.14, "outputPrice": 0.28, "cachedInputPrice": 0.028, "currency": "USD", "unit": "Per 1M tokens", "region": "Not separately published", "tier": "Standard published tier" }, "sourceId": "deepseek-pre-ipo-ascend-cluster", "source": { "id": "deepseek-pre-ipo-ascend-cluster", "title": "DeepSeek Infrastructure Architecture: 160,000-Accelerator Huawei Ascend CANN Cluster & Capital Disclosures", "publisher": "DeepSeek AI / Industry Regulatory Filings", "url": "https://deepseek.ai/corporate/infrastructure-2026", "published": "2026-10-10", "retrieved": "2026-10-12T07:00:00-04:00", "type": "Primary high-density AI infrastructure and corporate financial disclosure" } } ], "sourceId": "deepseek-pre-ipo-ascend-cluster", "sourceValidity": { "status": "primary-current", "label": "Primary source current", "primary": true, "urlValid": true, "retrievedAt": "2026-10-12T07:00:00-04:00", "ageDays": 7, "authority": "Primary high-density AI infrastructure and corporate financial disclosure", "publisher": "DeepSeek AI / Industry Regulatory Filings", "published": "2026-10-10" } }, { "key": "api|deepseek|v4-flash|usd|not-separately-published|standard-published-tier", "type": "api", "provider": "DeepSeek", "product": "V4 Flash", "currentTerms": { "inputPrice": 0.14, "outputPrice": 0.28, "cachedInputPrice": null, "currency": "USD", "unit": "Per 1M tokens", "region": "Not separately published", "tier": "Standard published tier" }, "effectiveFrom": null, "firstObserved": "2026-10-12", "lastObserved": "2026-10-19", "lastChanged": "2026-10-12", "observationCount": 2, "periodCount": 1, "periods": [ { "firstObserved": "2026-10-12", "lastObserved": "2026-10-19", "effectiveFrom": null, "terms": { "inputPrice": 0.14, "outputPrice": 0.28, "cachedInputPrice": null, "currency": "USD", "unit": "Per 1M tokens", "region": "Not separately published", "tier": "Standard published tier" }, "sourceId": "deepseek-v4-apr24", "source": { "id": "deepseek-v4-apr24", "title": "DeepSeek V4-Pro and V4-Flash release notes", "publisher": "DeepSeek", "url": "https://api-docs.deepseek.com/news/news260424/", "published": "2026-04-24", "retrieved": "2026-08-09T21:38:00-04:00", "type": "Primary model release documentation" } } ], "sourceId": "deepseek-v4-apr24", "sourceValidity": { "status": "stale", "label": "Source retrieval stale", "primary": true, "urlValid": true, "retrievedAt": "2026-08-09T21:38:00-04:00", "ageDays": 71, "authority": "Primary model release documentation", "publisher": "DeepSeek", "published": "2026-04-24" } }, { "key": "api|deepseek|v4-pro|usd|not-separately-published|standard-published-tier", "type": "api", "provider": "DeepSeek", "product": "V4 Pro", "currentTerms": { "inputPrice": 0.435, "outputPrice": 0.87, "cachedInputPrice": 0.003625, "currency": "USD", "unit": "Per 1M tokens", "region": "Not separately published", "tier": "Standard published tier" }, "effectiveFrom": null, "firstObserved": "2026-10-12", "lastObserved": "2026-10-19", "lastChanged": "2026-10-12", "observationCount": 2, "periodCount": 1, "periods": [ { "firstObserved": "2026-10-12", "lastObserved": "2026-10-19", "effectiveFrom": null, "terms": { "inputPrice": 0.435, "outputPrice": 0.87, "cachedInputPrice": 0.003625, "currency": "USD", "unit": "Per 1M tokens", "region": "Not separately published", "tier": "Standard published tier" }, "sourceId": "reuters-deepseek-v4-0813-aug13", "source": { "id": "reuters-deepseek-v4-0813-aug13", "title": "DeepSeek releases official V4 Pro model as it steps up expansion", "publisher": "Reuters", "url": "https://www.reuters.com/world/china/deepseek-releases-official-v4-pro-model-it-steps-up-expansion-2026-08-13/", "published": "2026-08-13", "retrieved": "2026-08-13T19:38:00-04:00", "type": "Current model release reporting" } } ], "sourceId": "reuters-deepseek-v4-0813-aug13", "sourceValidity": { "status": "stale", "label": "Source retrieval stale", "primary": false, "urlValid": true, "retrievedAt": "2026-08-13T19:38:00-04:00", "ageDays": 67, "authority": "Current model release reporting", "publisher": "Reuters", "published": "2026-08-13" } }, { "key": "api|google|gemini-2-5-pro-thinking-mode|usd|not-separately-published|standard-published-tier", "type": "api", "provider": "Google", "product": "Gemini 2.5 Pro (Thinking Mode)", "currentTerms": { "inputPrice": 1.25, "outputPrice": 5, "cachedInputPrice": 0.31, "currency": "USD", "unit": "Per 1M tokens", "region": "Not separately published", "tier": "Standard published tier" }, "effectiveFrom": null, "firstObserved": "2026-10-12", "lastObserved": "2026-10-19", "lastChanged": "2026-10-12", "observationCount": 2, "periodCount": 1, "periods": [ { "firstObserved": "2026-10-12", "lastObserved": "2026-10-19", "effectiveFrom": null, "terms": { "inputPrice": 1.25, "outputPrice": 5, "cachedInputPrice": 0.31, "currency": "USD", "unit": "Per 1M tokens", "region": "Not separately published", "tier": "Standard published tier" }, "sourceId": "google-deepmind-gemini-2-5-thinking-vertex", "source": { "id": "google-deepmind-gemini-2-5-thinking-vertex", "title": "Google Cloud Vertex AI: Gemini 2.5 Flash and Pro Thinking Mode with Native Recursive Code Execution", "publisher": "Google Cloud / Google DeepMind", "url": "https://cloud.google.com/blog/products/ai-machine-learning/gemini-2-5-thinking-vertex-ai", "published": "2026-10-06", "retrieved": "2026-10-08T09:00:00-04:00", "type": "Primary enterprise cloud AI platform release note" } } ], "sourceId": "google-deepmind-gemini-2-5-thinking-vertex", "sourceValidity": { "status": "primary-current", "label": "Primary source current", "primary": true, "urlValid": true, "retrievedAt": "2026-10-08T09:00:00-04:00", "ageDays": 11, "authority": "Primary enterprise cloud AI platform release note", "publisher": "Google Cloud / Google DeepMind", "published": "2026-10-06" } }, { "key": "api|google|gemini-2-5-pro-reasoner|usd|not-separately-published|standard-published-tier", "type": "api", "provider": "Google", "product": "Gemini 2.5 Pro Reasoner", "currentTerms": { "inputPrice": 1.25, "outputPrice": 5, "cachedInputPrice": 0.3125, "currency": "USD", "unit": "Per 1M tokens", "region": "Not separately published", "tier": "Standard published tier" }, "effectiveFrom": null, "firstObserved": "2026-10-12", "lastObserved": "2026-10-19", "lastChanged": "2026-10-12", "observationCount": 2, "periodCount": 1, "periods": [ { "firstObserved": "2026-10-12", "lastObserved": "2026-10-19", "effectiveFrom": null, "terms": { "inputPrice": 1.25, "outputPrice": 5, "cachedInputPrice": 0.3125, "currency": "USD", "unit": "Per 1M tokens", "region": "Not separately published", "tier": "Standard published tier" }, "sourceId": "google-gemini-2-5-pro-reasoner-paper", "source": { "id": "google-gemini-2-5-pro-reasoner-paper", "title": "Gemini 2.5 Pro Reasoner Technical Report: Step-Level Supervision, Monte Carlo Tree Search, and Robotics Foundation API", "publisher": "Google DeepMind", "url": "https://arxiv.org/abs/2609.12845", "published": "2026-09-21", "retrieved": "2026-09-21T09:00:00-04:00", "type": "Primary foundation model technical paper and developer release" } } ], "sourceId": "google-gemini-2-5-pro-reasoner-paper", "sourceValidity": { "status": "primary-current", "label": "Primary source current", "primary": true, "urlValid": true, "retrievedAt": "2026-09-21T09:00:00-04:00", "ageDays": 28, "authority": "Primary foundation model technical paper and developer release", "publisher": "Google DeepMind", "published": "2026-09-21" } }, { "key": "api|google-deepmind|gemini-3-8-flash-pro|usd|not-separately-published|standard-published-tier", "type": "api", "provider": "Google DeepMind", "product": "Gemini 3.8 (Flash & Pro)", "currentTerms": { "inputPrice": 0.075, "outputPrice": 0.3, "cachedInputPrice": 0.019, "currency": "USD", "unit": "Per 1M tokens", "region": "Not separately published", "tier": "Standard published tier" }, "effectiveFrom": null, "firstObserved": "2026-10-19", "lastObserved": "2026-10-19", "lastChanged": "2026-10-19", "observationCount": 1, "periodCount": 1, "periods": [ { "firstObserved": "2026-10-19", "lastObserved": "2026-10-19", "effectiveFrom": null, "terms": { "inputPrice": 0.075, "outputPrice": 0.3, "cachedInputPrice": 0.019, "currency": "USD", "unit": "Per 1M tokens", "region": "Not separately published", "tier": "Standard published tier" }, "sourceId": "google-gemini-38-multimodal-release", "source": { "id": "google-gemini-38-multimodal-release", "title": "Google DeepMind Gemini 3.8 (Flash & Pro) Architecture Whitepaper & System API Documentation", "publisher": "Google DeepMind", "url": "https://deepmind.google/technologies/gemini/", "published": "2026-10-15", "retrieved": "2026-10-19T07:00:00-04:00", "type": "Primary frontier multimodal model specification and 2M context API documentation" } } ], "sourceId": "google-gemini-38-multimodal-release", "sourceValidity": { "status": "primary-current", "label": "Primary source current", "primary": true, "urlValid": true, "retrievedAt": "2026-10-19T07:00:00-04:00", "ageDays": 0, "authority": "Primary frontier multimodal model specification and 2M context API documentation", "publisher": "Google DeepMind", "published": "2026-10-15" } }, { "key": "api|meta|llama-4-frontier-preview|usd|not-separately-published|standard-published-tier", "type": "api", "provider": "Meta", "product": "Llama 4 Frontier Preview", "currentTerms": { "inputPrice": 0.18, "outputPrice": 0.54, "cachedInputPrice": 0.045, "currency": "USD", "unit": "Per 1M tokens", "region": "Not separately published", "tier": "Standard published tier" }, "effectiveFrom": null, "firstObserved": "2026-10-12", "lastObserved": "2026-10-19", "lastChanged": "2026-10-12", "observationCount": 2, "periodCount": 1, "periods": [ { "firstObserved": "2026-10-12", "lastObserved": "2026-10-19", "effectiveFrom": null, "terms": { "inputPrice": 0.18, "outputPrice": 0.54, "cachedInputPrice": 0.045, "currency": "USD", "unit": "Per 1M tokens", "region": "Not separately published", "tier": "Standard published tier" }, "sourceId": "meta-llama4-frontier-preview-spec", "source": { "id": "meta-llama4-frontier-preview-spec", "title": "Llama 4 Frontier Preview Architecture Technical Report: 128-Expert Sparse MoE and Native MXFP4 Training at Scale", "publisher": "Meta AI Research", "url": "https://ai.meta.com/research/publications/llama-4-frontier-preview-report", "published": "2026-09-17", "retrieved": "2026-09-17T09:00:00-04:00", "type": "Primary foundation model technical report and repository release" } } ], "sourceId": "meta-llama4-frontier-preview-spec", "sourceValidity": { "status": "primary-current", "label": "Primary source current", "primary": true, "urlValid": true, "retrievedAt": "2026-09-17T09:00:00-04:00", "ageDays": 32, "authority": "Primary foundation model technical report and repository release", "publisher": "Meta AI Research", "published": "2026-09-17" } }, { "key": "api|meta|llama-4-scout-architecture-preview|usd|not-separately-published|standard-published-tier", "type": "api", "provider": "Meta", "product": "Llama 4 Scout (Architecture Preview)", "currentTerms": { "inputPrice": 0.2, "outputPrice": 0.6, "cachedInputPrice": 0.05, "currency": "USD", "unit": "Per 1M tokens", "region": "Not separately published", "tier": "Standard published tier" }, "effectiveFrom": null, "firstObserved": "2026-10-12", "lastObserved": "2026-10-19", "lastChanged": "2026-10-12", "observationCount": 2, "periodCount": 1, "periods": [ { "firstObserved": "2026-10-12", "lastObserved": "2026-10-19", "effectiveFrom": null, "terms": { "inputPrice": 0.2, "outputPrice": 0.6, "cachedInputPrice": 0.05, "currency": "USD", "unit": "Per 1M tokens", "region": "Not separately published", "tier": "Standard published tier" }, "sourceId": "meta-llama-4-speculative-decoding-preview", "source": { "id": "meta-llama-4-speculative-decoding-preview", "title": "Meta AI Engineering: Llama 4 Architectural Blueprint, Native Speculative Decoding, and MoE Routing at Scale", "publisher": "Meta AI / Fundamental AI Research", "url": "https://ai.meta.com/blog/llama-4-architecture-preview/", "published": "2026-10-07", "retrieved": "2026-10-08T09:00:00-04:00", "type": "Primary open-weight architectural paper and release specification" } } ], "sourceId": "meta-llama-4-speculative-decoding-preview", "sourceValidity": { "status": "primary-current", "label": "Primary source current", "primary": true, "urlValid": true, "retrievedAt": "2026-10-08T09:00:00-04:00", "ageDays": 11, "authority": "Primary open-weight architectural paper and release specification", "publisher": "Meta AI / Fundamental AI Research", "published": "2026-10-07" } }, { "key": "api|mistral-ai|mistral-large-3|usd|not-separately-published|standard-published-tier", "type": "api", "provider": "Mistral AI", "product": "Mistral Large 3", "currentTerms": { "inputPrice": 0.2, "outputPrice": 0.6, "cachedInputPrice": 0.05, "currency": "USD", "unit": "Per 1M tokens", "region": "Not separately published", "tier": "Standard published tier" }, "effectiveFrom": null, "firstObserved": "2026-10-12", "lastObserved": "2026-10-19", "lastChanged": "2026-10-12", "observationCount": 2, "periodCount": 1, "periods": [ { "firstObserved": "2026-10-12", "lastObserved": "2026-10-19", "effectiveFrom": null, "terms": { "inputPrice": 0.2, "outputPrice": 0.6, "cachedInputPrice": 0.05, "currency": "USD", "unit": "Per 1M tokens", "region": "Not separately published", "tier": "Standard published tier" }, "sourceId": "mistral-large-3-hybrid-ssm-paper", "source": { "id": "mistral-large-3-hybrid-ssm-paper", "title": "Mistral Large 3: Dense 145B Hybrid Selective State-Space and Transformer Model Architecture", "publisher": "Mistral AI Research", "url": "https://mistral.ai/news/mistral-large-3/", "published": "2026-09-21", "retrieved": "2026-09-21T09:00:00-04:00", "type": "Primary model technical release and architecture whitepaper" } } ], "sourceId": "mistral-large-3-hybrid-ssm-paper", "sourceValidity": { "status": "primary-current", "label": "Primary source current", "primary": true, "urlValid": true, "retrievedAt": "2026-09-21T09:00:00-04:00", "ageDays": 28, "authority": "Primary model technical release and architecture whitepaper", "publisher": "Mistral AI Research", "published": "2026-09-21" } }, { "key": "api|moonshot-ai|kimi-k3|cny|not-separately-published|standard-published-tier", "type": "api", "provider": "Moonshot AI", "product": "Kimi K3", "currentTerms": { "inputPrice": 20, "outputPrice": 100, "cachedInputPrice": 2, "currency": "CNY", "unit": "Per 1M tokens", "region": "Not separately published", "tier": "Standard published tier" }, "effectiveFrom": null, "firstObserved": "2026-10-12", "lastObserved": "2026-10-19", "lastChanged": "2026-10-12", "observationCount": 2, "periodCount": 1, "periods": [ { "firstObserved": "2026-10-12", "lastObserved": "2026-10-19", "effectiveFrom": null, "terms": { "inputPrice": 20, "outputPrice": 100, "cachedInputPrice": 2, "currency": "CNY", "unit": "Per 1M tokens", "region": "Not separately published", "tier": "Standard published tier" }, "sourceId": "moonshot-kimi-k3", "source": { "id": "moonshot-kimi-k3", "title": "Kimi API platform — Kimi K3 model and pricing", "publisher": "Moonshot AI", "url": "https://platform.moonshot.cn/", "published": "Current page", "retrieved": "2026-07-20T20:00:00-04:00", "type": "Primary" } } ], "sourceId": "moonshot-kimi-k3", "sourceValidity": { "status": "stale", "label": "Source retrieval stale", "primary": true, "urlValid": true, "retrievedAt": "2026-07-20T20:00:00-04:00", "ageDays": 91, "authority": "Primary", "publisher": "Moonshot AI", "published": "Current page" } }, { "key": "api|openai|gpt-5-6-luna|usd|not-separately-published|standard-published-tier", "type": "api", "provider": "OpenAI", "product": "GPT-5.6 Luna", "currentTerms": { "inputPrice": 1, "outputPrice": 6, "cachedInputPrice": null, "currency": "USD", "unit": "Per 1M tokens", "region": "Not separately published", "tier": "Standard published tier" }, "effectiveFrom": null, "firstObserved": "2026-10-12", "lastObserved": "2026-10-19", "lastChanged": "2026-10-12", "observationCount": 2, "periodCount": 1, "periods": [ { "firstObserved": "2026-10-12", "lastObserved": "2026-10-19", "effectiveFrom": null, "terms": { "inputPrice": 1, "outputPrice": 6, "cachedInputPrice": null, "currency": "USD", "unit": "Per 1M tokens", "region": "Not separately published", "tier": "Standard published tier" }, "sourceId": "openai-gpt56", "source": { "id": "openai-gpt56", "title": "Introducing GPT-5.6", "publisher": "OpenAI", "url": "https://openai.com/index/gpt-5-6/", "published": "2026-07-09", "retrieved": "2026-07-20T19:20:00-04:00", "type": "Primary" } } ], "sourceId": "openai-gpt56", "sourceValidity": { "status": "stale", "label": "Source retrieval stale", "primary": true, "urlValid": true, "retrievedAt": "2026-07-20T19:20:00-04:00", "ageDays": 91, "authority": "Primary", "publisher": "OpenAI", "published": "2026-07-09" } }, { "key": "api|openai|gpt-5-6-sol|usd|not-separately-published|standard-published-tier", "type": "api", "provider": "OpenAI", "product": "GPT-5.6 Sol", "currentTerms": { "inputPrice": 5, "outputPrice": 30, "cachedInputPrice": null, "currency": "USD", "unit": "Per 1M tokens", "region": "Not separately published", "tier": "Standard published tier" }, "effectiveFrom": null, "firstObserved": "2026-10-12", "lastObserved": "2026-10-19", "lastChanged": "2026-10-12", "observationCount": 2, "periodCount": 1, "periods": [ { "firstObserved": "2026-10-12", "lastObserved": "2026-10-19", "effectiveFrom": null, "terms": { "inputPrice": 5, "outputPrice": 30, "cachedInputPrice": null, "currency": "USD", "unit": "Per 1M tokens", "region": "Not separately published", "tier": "Standard published tier" }, "sourceId": "openai-gpt56", "source": { "id": "openai-gpt56", "title": "Introducing GPT-5.6", "publisher": "OpenAI", "url": "https://openai.com/index/gpt-5-6/", "published": "2026-07-09", "retrieved": "2026-07-20T19:20:00-04:00", "type": "Primary" } } ], "sourceId": "openai-gpt56", "sourceValidity": { "status": "stale", "label": "Source retrieval stale", "primary": true, "urlValid": true, "retrievedAt": "2026-07-20T19:20:00-04:00", "ageDays": 91, "authority": "Primary", "publisher": "OpenAI", "published": "2026-07-09" } }, { "key": "api|openai|gpt-5-6-terra|usd|not-separately-published|standard-published-tier", "type": "api", "provider": "OpenAI", "product": "GPT-5.6 Terra", "currentTerms": { "inputPrice": 2.5, "outputPrice": 15, "cachedInputPrice": null, "currency": "USD", "unit": "Per 1M tokens", "region": "Not separately published", "tier": "Standard published tier" }, "effectiveFrom": null, "firstObserved": "2026-10-12", "lastObserved": "2026-10-19", "lastChanged": "2026-10-12", "observationCount": 2, "periodCount": 1, "periods": [ { "firstObserved": "2026-10-12", "lastObserved": "2026-10-19", "effectiveFrom": null, "terms": { "inputPrice": 2.5, "outputPrice": 15, "cachedInputPrice": null, "currency": "USD", "unit": "Per 1M tokens", "region": "Not separately published", "tier": "Standard published tier" }, "sourceId": "openai-gpt56", "source": { "id": "openai-gpt56", "title": "Introducing GPT-5.6", "publisher": "OpenAI", "url": "https://openai.com/index/gpt-5-6/", "published": "2026-07-09", "retrieved": "2026-07-20T19:20:00-04:00", "type": "Primary" } } ], "sourceId": "openai-gpt56", "sourceValidity": { "status": "stale", "label": "Source retrieval stale", "primary": true, "urlValid": true, "retrievedAt": "2026-07-20T19:20:00-04:00", "ageDays": 91, "authority": "Primary", "publisher": "OpenAI", "published": "2026-07-09" } }, { "key": "api|openai|gpt-6-sol-luna|usd|not-separately-published|standard-published-tier", "type": "api", "provider": "OpenAI", "product": "GPT-6 (Sol & Luna)", "currentTerms": { "inputPrice": 0.35, "outputPrice": 1.4, "cachedInputPrice": 0.08, "currency": "USD", "unit": "Per 1M tokens", "region": "Not separately published", "tier": "Standard published tier" }, "effectiveFrom": null, "firstObserved": "2026-10-12", "lastObserved": "2026-10-19", "lastChanged": "2026-10-12", "observationCount": 2, "periodCount": 1, "periods": [ { "firstObserved": "2026-10-12", "lastObserved": "2026-10-19", "effectiveFrom": null, "terms": { "inputPrice": 0.35, "outputPrice": 1.4, "cachedInputPrice": 0.08, "currency": "USD", "unit": "Per 1M tokens", "region": "Not separately published", "tier": "Standard published tier" }, "sourceId": "openai-gpt6-ga-whitepaper", "source": { "id": "openai-gpt6-ga-whitepaper", "title": "OpenAI GPT-6 (Sol & Luna) Dual-Engine Production Deployment Specification", "publisher": "OpenAI", "url": "https://openai.com/index/intelligent-ui", "published": "2026-10-14", "retrieved": "2026-10-19T07:00:00-04:00", "type": "Primary frontier model platform and dual-engine architecture release" } } ], "sourceId": "openai-gpt6-ga-whitepaper", "sourceValidity": { "status": "primary-current", "label": "Primary source current", "primary": true, "urlValid": true, "retrievedAt": "2026-10-19T07:00:00-04:00", "ageDays": 0, "authority": "Primary frontier model platform and dual-engine architecture release", "publisher": "OpenAI", "published": "2026-10-14" } }, { "key": "api|openai|o3-mini-high-effort|usd|not-separately-published|standard-published-tier", "type": "api", "provider": "OpenAI", "product": "o3-mini (High Effort)", "currentTerms": { "inputPrice": 0.4, "outputPrice": 1.6, "cachedInputPrice": 0.2, "currency": "USD", "unit": "Per 1M tokens", "region": "Not separately published", "tier": "Standard published tier" }, "effectiveFrom": null, "firstObserved": "2026-10-12", "lastObserved": "2026-10-19", "lastChanged": "2026-10-12", "observationCount": 2, "periodCount": 1, "periods": [ { "firstObserved": "2026-10-12", "lastObserved": "2026-10-19", "effectiveFrom": null, "terms": { "inputPrice": 0.4, "outputPrice": 1.6, "cachedInputPrice": 0.2, "currency": "USD", "unit": "Per 1M tokens", "region": "Not separately published", "tier": "Standard published tier" }, "sourceId": "openai-o3-mini-system-card", "source": { "id": "openai-o3-mini-system-card", "title": "OpenAI o3-mini System Card & Technical Documentation: Controllable Test-Time Compute, Mathematical Reasoning, and MicroVM Sandboxing", "publisher": "OpenAI", "url": "https://openai.com/index/openai-o3-mini/", "published": "2026-09-22", "retrieved": "2026-09-22T09:00:00-04:00", "type": "Primary foundation model release and developer system card" } } ], "sourceId": "openai-o3-mini-system-card", "sourceValidity": { "status": "primary-current", "label": "Primary source current", "primary": true, "urlValid": true, "retrievedAt": "2026-09-22T09:00:00-04:00", "ageDays": 27, "authority": "Primary foundation model release and developer system card", "publisher": "OpenAI", "published": "2026-09-22" } }, { "key": "api|openai|o3-mini-medium-effort|usd|not-separately-published|standard-published-tier", "type": "api", "provider": "OpenAI", "product": "o3-mini (Medium Effort)", "currentTerms": { "inputPrice": 0.4, "outputPrice": 1.6, "cachedInputPrice": 0.2, "currency": "USD", "unit": "Per 1M tokens", "region": "Not separately published", "tier": "Standard published tier" }, "effectiveFrom": null, "firstObserved": "2026-10-12", "lastObserved": "2026-10-19", "lastChanged": "2026-10-12", "observationCount": 2, "periodCount": 1, "periods": [ { "firstObserved": "2026-10-12", "lastObserved": "2026-10-19", "effectiveFrom": null, "terms": { "inputPrice": 0.4, "outputPrice": 1.6, "cachedInputPrice": 0.2, "currency": "USD", "unit": "Per 1M tokens", "region": "Not separately published", "tier": "Standard published tier" }, "sourceId": "openai-o3-mini-system-card", "source": { "id": "openai-o3-mini-system-card", "title": "OpenAI o3-mini System Card & Technical Documentation: Controllable Test-Time Compute, Mathematical Reasoning, and MicroVM Sandboxing", "publisher": "OpenAI", "url": "https://openai.com/index/openai-o3-mini/", "published": "2026-09-22", "retrieved": "2026-09-22T09:00:00-04:00", "type": "Primary foundation model release and developer system card" } } ], "sourceId": "openai-o3-mini-system-card", "sourceValidity": { "status": "primary-current", "label": "Primary source current", "primary": true, "urlValid": true, "retrievedAt": "2026-09-22T09:00:00-04:00", "ageDays": 27, "authority": "Primary foundation model release and developer system card", "publisher": "OpenAI", "published": "2026-09-22" } }, { "key": "api|openai|openai-o3-pro|usd|not-separately-published|standard-published-tier", "type": "api", "provider": "OpenAI", "product": "OpenAI o3-pro", "currentTerms": { "inputPrice": 15, "outputPrice": 60, "cachedInputPrice": 3.75, "currency": "USD", "unit": "Per 1M tokens", "region": "Not separately published", "tier": "Standard published tier" }, "effectiveFrom": null, "firstObserved": "2026-10-12", "lastObserved": "2026-10-19", "lastChanged": "2026-10-12", "observationCount": 2, "periodCount": 1, "periods": [ { "firstObserved": "2026-10-12", "lastObserved": "2026-10-19", "effectiveFrom": null, "terms": { "inputPrice": 15, "outputPrice": 60, "cachedInputPrice": 3.75, "currency": "USD", "unit": "Per 1M tokens", "region": "Not separately published", "tier": "Standard published tier" }, "sourceId": "openai-o3-pro-system-card", "source": { "id": "openai-o3-pro-system-card", "title": "OpenAI o3-pro System Card: Frontier Reasoning, Test-Time Tree Search with Native Multi-Step Tool Execution, and Autonomous Coding Evaluations", "publisher": "OpenAI", "url": "https://openai.com/index/o3-pro-system-card", "published": "2026-09-28", "retrieved": "2026-09-28T09:00:00-04:00", "type": "Primary foundation model system card and technical specification" } } ], "sourceId": "openai-o3-pro-system-card", "sourceValidity": { "status": "primary-current", "label": "Primary source current", "primary": true, "urlValid": true, "retrievedAt": "2026-09-28T09:00:00-04:00", "ageDays": 21, "authority": "Primary foundation model system card and technical specification", "publisher": "OpenAI", "published": "2026-09-28" } }, { "key": "api|openai|openai-operator-web-agent|usd|not-separately-published|standard-published-tier", "type": "api", "provider": "OpenAI", "product": "OpenAI Operator (Web Agent)", "currentTerms": { "inputPrice": 5, "outputPrice": 20, "cachedInputPrice": 1.25, "currency": "USD", "unit": "Per 1M tokens", "region": "Not separately published", "tier": "Standard published tier" }, "effectiveFrom": null, "firstObserved": "2026-10-12", "lastObserved": "2026-10-19", "lastChanged": "2026-10-12", "observationCount": 2, "periodCount": 1, "periods": [ { "firstObserved": "2026-10-12", "lastObserved": "2026-10-19", "effectiveFrom": null, "terms": { "inputPrice": 5, "outputPrice": 20, "cachedInputPrice": 1.25, "currency": "USD", "unit": "Per 1M tokens", "region": "Not separately published", "tier": "Standard published tier" }, "sourceId": "openai-operator-sandbox-architecture", "source": { "id": "openai-operator-sandbox-architecture", "title": "OpenAI Operator Architecture: MicroVM Hardware Isolation, Permission Gating, and Dual-Loop Verification for Web Agents", "publisher": "OpenAI", "url": "https://openai.com/index/introducing-operator/", "published": "2026-10-07", "retrieved": "2026-10-08T09:00:00-04:00", "type": "Primary foundation model and autonomous agent technical specification" } } ], "sourceId": "openai-operator-sandbox-architecture", "sourceValidity": { "status": "primary-current", "label": "Primary source current", "primary": true, "urlValid": true, "retrievedAt": "2026-10-08T09:00:00-04:00", "ageDays": 11, "authority": "Primary foundation model and autonomous agent technical specification", "publisher": "OpenAI", "published": "2026-10-07" } }, { "key": "api|stepfun|step-3-7-flash|cny|not-separately-published|standard-published-tier", "type": "api", "provider": "StepFun", "product": "Step 3.7 Flash", "currentTerms": { "inputPrice": 1.35, "outputPrice": 8.1, "cachedInputPrice": 0.27, "currency": "CNY", "unit": "Per 1M tokens", "region": "Not separately published", "tier": "Standard published tier" }, "effectiveFrom": null, "firstObserved": "2026-10-12", "lastObserved": "2026-10-19", "lastChanged": "2026-10-12", "observationCount": 2, "periodCount": 1, "periods": [ { "firstObserved": "2026-10-12", "lastObserved": "2026-10-19", "effectiveFrom": null, "terms": { "inputPrice": 1.35, "outputPrice": 8.1, "cachedInputPrice": 0.27, "currency": "CNY", "unit": "Per 1M tokens", "region": "Not separately published", "tier": "Standard published tier" }, "sourceId": "stepfun-37-flash", "source": { "id": "stepfun-37-flash", "title": "StepFun — Step 3.7 Flash pricing and model documentation", "publisher": "StepFun", "url": "https://platform.stepfun.com/docs/zh/guides/pricing/details", "published": "Current page", "retrieved": "2026-07-20T20:00:00-04:00", "type": "Primary" } } ], "sourceId": "stepfun-37-flash", "sourceValidity": { "status": "stale", "label": "Source retrieval stale", "primary": true, "urlValid": true, "retrievedAt": "2026-07-20T20:00:00-04:00", "ageDays": 91, "authority": "Primary", "publisher": "StepFun", "published": "Current page" } }, { "key": "api|xai|grok-4-5|usd|not-separately-published|standard-published-tier", "type": "api", "provider": "xAI", "product": "Grok 4.5", "currentTerms": { "inputPrice": 2, "outputPrice": 6, "cachedInputPrice": 0.3, "currency": "USD", "unit": "Per 1M tokens", "region": "Not separately published", "tier": "Standard published tier" }, "effectiveFrom": null, "firstObserved": "2026-10-12", "lastObserved": "2026-10-19", "lastChanged": "2026-10-12", "observationCount": 2, "periodCount": 1, "periods": [ { "firstObserved": "2026-10-12", "lastObserved": "2026-10-19", "effectiveFrom": null, "terms": { "inputPrice": 2, "outputPrice": 6, "cachedInputPrice": 0.3, "currency": "USD", "unit": "Per 1M tokens", "region": "Not separately published", "tier": "Standard published tier" }, "sourceId": "xai-pricing", "source": { "id": "xai-pricing", "title": "xAI API pricing", "publisher": "xAI", "url": "https://docs.x.ai/developers/pricing", "published": "2026-07", "retrieved": "2026-07-20T20:00:00-04:00", "type": "Primary" } } ], "sourceId": "xai-pricing", "sourceValidity": { "status": "stale", "label": "Source retrieval stale", "primary": true, "urlValid": true, "retrievedAt": "2026-07-20T20:00:00-04:00", "ageDays": 91, "authority": "Primary", "publisher": "xAI", "published": "2026-07" } }, { "key": "free-tier|openai|chatgpt-free-go-plus-pro-business-enterprise|not-separately-published", "type": "freeTier", "provider": "OpenAI", "product": "ChatGPT Free / Go / Plus / Pro / Business / Enterprise", "currentTerms": { "price": "Free tier included in plan family", "currency": "Not separately published", "region": "Not separately published", "limit": "Not separately quantified in this edition", "eligibility": "Not separately published", "feature": "Plus includes GPT-5.6 advanced reasoning; exact prices can vary by region.", "notes": "Derived from the preserved plan record until a dedicated provider limit is captured." }, "effectiveFrom": null, "firstObserved": "2026-10-12", "lastObserved": "2026-10-19", "lastChanged": "2026-10-12", "observationCount": 2, "periodCount": 1, "periods": [ { "firstObserved": "2026-10-12", "lastObserved": "2026-10-19", "effectiveFrom": null, "terms": { "price": "Free tier included in plan family", "currency": "Not separately published", "region": "Not separately published", "limit": "Not separately quantified in this edition", "eligibility": "Not separately published", "feature": "Plus includes GPT-5.6 advanced reasoning; exact prices can vary by region.", "notes": "Derived from the preserved plan record until a dedicated provider limit is captured." }, "sourceId": "openai-chatgpt-pricing", "source": { "id": "openai-chatgpt-pricing", "title": "ChatGPT Plans", "publisher": "OpenAI", "url": "https://openai.com/chatgpt/pricing/", "published": "Current page", "retrieved": "2026-07-20T20:00:00-04:00", "type": "Primary" } } ], "sourceId": "openai-chatgpt-pricing", "sourceValidity": { "status": "stale", "label": "Source retrieval stale", "primary": true, "urlValid": true, "retrievedAt": "2026-07-20T20:00:00-04:00", "ageDays": 91, "authority": "Primary", "publisher": "OpenAI", "published": "Current page" } }, { "key": "free-tier|perplexity|standard-pro-max-education-pro|not-separately-published", "type": "freeTier", "provider": "Perplexity", "product": "Standard / Pro / Max / Education Pro", "currentTerms": { "price": "Free tier included in plan family", "currency": "Not separately published", "region": "Not separately published", "limit": "Not separately quantified in this edition", "eligibility": "Not separately published", "feature": "Individual plans range from free search to higher-limit Pro and Max tiers; eligibility and regional terms vary.", "notes": "Derived from the preserved plan record until a dedicated provider limit is captured." }, "effectiveFrom": null, "firstObserved": "2026-10-12", "lastObserved": "2026-10-19", "lastChanged": "2026-10-12", "observationCount": 2, "periodCount": 1, "periods": [ { "firstObserved": "2026-10-12", "lastObserved": "2026-10-19", "effectiveFrom": null, "terms": { "price": "Free tier included in plan family", "currency": "Not separately published", "region": "Not separately published", "limit": "Not separately quantified in this edition", "eligibility": "Not separately published", "feature": "Individual plans range from free search to higher-limit Pro and Max tiers; eligibility and regional terms vary.", "notes": "Derived from the preserved plan record until a dedicated provider limit is captured." }, "sourceId": "perplexity-plans", "source": { "id": "perplexity-plans", "title": "Which Perplexity subscription plan is right for you?", "publisher": "Perplexity Support", "url": "https://www.perplexity.ai/help-center/en/articles/11187416-which-perplexity-subscription-plan-is-right-for-you", "published": "Updated 2026-07-22", "retrieved": "2026-07-23T07:11:00-04:00", "type": "Primary plan documentation" } } ], "sourceId": "perplexity-plans", "sourceValidity": { "status": "stale", "label": "Source retrieval stale", "primary": true, "urlValid": true, "retrievedAt": "2026-07-23T07:11:00-04:00", "ageDays": 88, "authority": "Primary plan documentation", "publisher": "Perplexity Support", "published": "Updated 2026-07-22" } }, { "key": "plan|anthropic|claude-max|not-separately-published", "type": "plan", "provider": "Anthropic", "product": "Claude Max", "currentTerms": { "price": "Verify current regional checkout", "currency": "Not separately published", "billingPeriod": "Not separately published", "region": "Not separately published", "feature": "Higher individual usage tiers for heavier Claude users where available." }, "effectiveFrom": null, "firstObserved": "2026-10-12", "lastObserved": "2026-10-19", "lastChanged": "2026-10-12", "observationCount": 2, "periodCount": 1, "periods": [ { "firstObserved": "2026-10-12", "lastObserved": "2026-10-19", "effectiveFrom": null, "terms": { "price": "Verify current regional checkout", "currency": "Not separately published", "billingPeriod": "Not separately published", "region": "Not separately published", "feature": "Higher individual usage tiers for heavier Claude users where available." }, "sourceId": "anthropic-claude", "source": { "id": "anthropic-claude", "title": "Claude product overview", "publisher": "Anthropic", "url": "https://claude.com/", "published": "Current page", "retrieved": "2026-07-20T20:00:00-04:00", "type": "Primary" } } ], "sourceId": "anthropic-claude", "sourceValidity": { "status": "stale", "label": "Source retrieval stale", "primary": true, "urlValid": true, "retrievedAt": "2026-07-20T20:00:00-04:00", "ageDays": 91, "authority": "Primary", "publisher": "Anthropic", "published": "Current page" } }, { "key": "plan|anthropic|claude-pro|not-separately-published", "type": "plan", "provider": "Anthropic", "product": "Claude Pro", "currentTerms": { "price": "Verify current regional checkout", "currency": "Not separately published", "billingPeriod": "Not separately published", "region": "Not separately published", "feature": "Individual paid tier with higher usage and access to current Claude models where available." }, "effectiveFrom": null, "firstObserved": "2026-10-12", "lastObserved": "2026-10-19", "lastChanged": "2026-10-12", "observationCount": 2, "periodCount": 1, "periods": [ { "firstObserved": "2026-10-12", "lastObserved": "2026-10-19", "effectiveFrom": null, "terms": { "price": "Verify current regional checkout", "currency": "Not separately published", "billingPeriod": "Not separately published", "region": "Not separately published", "feature": "Individual paid tier with higher usage and access to current Claude models where available." }, "sourceId": "anthropic-claude", "source": { "id": "anthropic-claude", "title": "Claude product overview", "publisher": "Anthropic", "url": "https://claude.com/", "published": "Current page", "retrieved": "2026-07-20T20:00:00-04:00", "type": "Primary" } } ], "sourceId": "anthropic-claude", "sourceValidity": { "status": "stale", "label": "Source retrieval stale", "primary": true, "urlValid": true, "retrievedAt": "2026-07-20T20:00:00-04:00", "ageDays": 91, "authority": "Primary", "publisher": "Anthropic", "published": "Current page" } }, { "key": "plan|google|google-ai-plus|not-separately-published", "type": "plan", "provider": "Google", "product": "Google AI Plus", "currentTerms": { "price": "$4.99 USD/month", "currency": "USD", "billingPeriod": "Monthly", "region": "Not separately published", "feature": "400GB storage and expanded Gemini access; availability and local tax vary." }, "effectiveFrom": null, "firstObserved": "2026-10-12", "lastObserved": "2026-10-19", "lastChanged": "2026-10-12", "observationCount": 2, "periodCount": 1, "periods": [ { "firstObserved": "2026-10-12", "lastObserved": "2026-10-19", "effectiveFrom": null, "terms": { "price": "$4.99 USD/month", "currency": "USD", "billingPeriod": "Monthly", "region": "Not separately published", "feature": "400GB storage and expanded Gemini access; availability and local tax vary." }, "sourceId": "google-ai-plans", "source": { "id": "google-ai-plans", "title": "Google AI plans", "publisher": "Google", "url": "https://one.google.com/about/google-ai-plans/", "published": "Current page", "retrieved": "2026-07-20T20:00:00-04:00", "type": "Primary" } } ], "sourceId": "google-ai-plans", "sourceValidity": { "status": "stale", "label": "Source retrieval stale", "primary": true, "urlValid": true, "retrievedAt": "2026-07-20T20:00:00-04:00", "ageDays": 91, "authority": "Primary", "publisher": "Google", "published": "Current page" } }, { "key": "plan|google|google-ai-pro|not-separately-published", "type": "plan", "provider": "Google", "product": "Google AI Pro", "currentTerms": { "price": "$19.99 USD/month", "currency": "USD", "billingPeriod": "Monthly", "region": "Not separately published", "feature": "5TB storage and higher access to Gemini tools; availability and local tax vary." }, "effectiveFrom": null, "firstObserved": "2026-10-12", "lastObserved": "2026-10-19", "lastChanged": "2026-10-12", "observationCount": 2, "periodCount": 1, "periods": [ { "firstObserved": "2026-10-12", "lastObserved": "2026-10-19", "effectiveFrom": null, "terms": { "price": "$19.99 USD/month", "currency": "USD", "billingPeriod": "Monthly", "region": "Not separately published", "feature": "5TB storage and higher access to Gemini tools; availability and local tax vary." }, "sourceId": "google-ai-plans", "source": { "id": "google-ai-plans", "title": "Google AI plans", "publisher": "Google", "url": "https://one.google.com/about/google-ai-plans/", "published": "Current page", "retrieved": "2026-07-20T20:00:00-04:00", "type": "Primary" } } ], "sourceId": "google-ai-plans", "sourceValidity": { "status": "stale", "label": "Source retrieval stale", "primary": true, "urlValid": true, "retrievedAt": "2026-07-20T20:00:00-04:00", "ageDays": 91, "authority": "Primary", "publisher": "Google", "published": "Current page" } }, { "key": "plan|microsoft|copilot-pro|not-separately-published", "type": "plan", "provider": "Microsoft", "product": "Copilot Pro", "currentTerms": { "price": "$20 USD/user/month", "currency": "USD", "billingPeriod": "Monthly", "region": "Not separately published", "feature": "Preferred access to advanced models and Copilot features in supported Microsoft 365 apps." }, "effectiveFrom": null, "firstObserved": "2026-10-12", "lastObserved": "2026-10-19", "lastChanged": "2026-10-12", "observationCount": 2, "periodCount": 1, "periods": [ { "firstObserved": "2026-10-12", "lastObserved": "2026-10-19", "effectiveFrom": null, "terms": { "price": "$20 USD/user/month", "currency": "USD", "billingPeriod": "Monthly", "region": "Not separately published", "feature": "Preferred access to advanced models and Copilot features in supported Microsoft 365 apps." }, "sourceId": "microsoft-copilot-pro", "source": { "id": "microsoft-copilot-pro", "title": "Microsoft Copilot Pro plan and pricing", "publisher": "Microsoft", "url": "https://www.microsoft.com/en-us/store/b/copilotpro", "published": "Current page", "retrieved": "2026-07-23T07:09:00-04:00", "type": "Primary plan page" } } ], "sourceId": "microsoft-copilot-pro", "sourceValidity": { "status": "stale", "label": "Source retrieval stale", "primary": true, "urlValid": true, "retrievedAt": "2026-07-23T07:09:00-04:00", "ageDays": 88, "authority": "Primary plan page", "publisher": "Microsoft", "published": "Current page" } }, { "key": "plan|openai|chatgpt-free-go-plus-pro-business-enterprise|not-separately-published", "type": "plan", "provider": "OpenAI", "product": "ChatGPT Free / Go / Plus / Pro / Business / Enterprise", "currentTerms": { "price": "Regional price shown at checkout", "currency": "Not separately published", "billingPeriod": "Not separately published", "region": "Not separately published", "feature": "Plus includes GPT-5.6 advanced reasoning; exact prices can vary by region." }, "effectiveFrom": null, "firstObserved": "2026-10-12", "lastObserved": "2026-10-19", "lastChanged": "2026-10-12", "observationCount": 2, "periodCount": 1, "periods": [ { "firstObserved": "2026-10-12", "lastObserved": "2026-10-19", "effectiveFrom": null, "terms": { "price": "Regional price shown at checkout", "currency": "Not separately published", "billingPeriod": "Not separately published", "region": "Not separately published", "feature": "Plus includes GPT-5.6 advanced reasoning; exact prices can vary by region." }, "sourceId": "openai-chatgpt-pricing", "source": { "id": "openai-chatgpt-pricing", "title": "ChatGPT Plans", "publisher": "OpenAI", "url": "https://openai.com/chatgpt/pricing/", "published": "Current page", "retrieved": "2026-07-20T20:00:00-04:00", "type": "Primary" } } ], "sourceId": "openai-chatgpt-pricing", "sourceValidity": { "status": "stale", "label": "Source retrieval stale", "primary": true, "urlValid": true, "retrievedAt": "2026-07-20T20:00:00-04:00", "ageDays": 91, "authority": "Primary", "publisher": "OpenAI", "published": "Current page" } }, { "key": "plan|perplexity|standard-pro-max-education-pro|not-separately-published", "type": "plan", "provider": "Perplexity", "product": "Standard / Pro / Max / Education Pro", "currentTerms": { "price": "Education Pro: $10/month with verification; verify other paid tiers", "currency": "USD", "billingPeriod": "Monthly", "region": "Not separately published", "feature": "Individual plans range from free search to higher-limit Pro and Max tiers; eligibility and regional terms vary." }, "effectiveFrom": null, "firstObserved": "2026-10-12", "lastObserved": "2026-10-19", "lastChanged": "2026-10-12", "observationCount": 2, "periodCount": 1, "periods": [ { "firstObserved": "2026-10-12", "lastObserved": "2026-10-19", "effectiveFrom": null, "terms": { "price": "Education Pro: $10/month with verification; verify other paid tiers", "currency": "USD", "billingPeriod": "Monthly", "region": "Not separately published", "feature": "Individual plans range from free search to higher-limit Pro and Max tiers; eligibility and regional terms vary." }, "sourceId": "perplexity-plans", "source": { "id": "perplexity-plans", "title": "Which Perplexity subscription plan is right for you?", "publisher": "Perplexity Support", "url": "https://www.perplexity.ai/help-center/en/articles/11187416-which-perplexity-subscription-plan-is-right-for-you", "published": "Updated 2026-07-22", "retrieved": "2026-07-23T07:11:00-04:00", "type": "Primary plan documentation" } } ], "sourceId": "perplexity-plans", "sourceValidity": { "status": "stale", "label": "Source retrieval stale", "primary": true, "urlValid": true, "retrievedAt": "2026-07-23T07:11:00-04:00", "ageDays": 88, "authority": "Primary plan documentation", "publisher": "Perplexity Support", "published": "Updated 2026-07-22" } }, { "key": "plan|xai|grok-premium-plans|not-separately-published", "type": "plan", "provider": "xAI", "product": "Grok premium plans", "currentTerms": { "price": "Verify current checkout", "currency": "Not separately published", "billingPeriod": "Not separately published", "region": "Not separately published", "feature": "Plan names, billing terms and regional availability can change; use the official product page." }, "effectiveFrom": null, "firstObserved": "2026-10-12", "lastObserved": "2026-10-19", "lastChanged": "2026-10-12", "observationCount": 2, "periodCount": 1, "periods": [ { "firstObserved": "2026-10-12", "lastObserved": "2026-10-19", "effectiveFrom": null, "terms": { "price": "Verify current checkout", "currency": "Not separately published", "billingPeriod": "Not separately published", "region": "Not separately published", "feature": "Plan names, billing terms and regional availability can change; use the official product page." }, "sourceId": "xai-grok-plans", "source": { "id": "xai-grok-plans", "title": "Grok product and plan page", "publisher": "xAI", "url": "https://grok.com/", "published": "Current page", "retrieved": "2026-07-23T07:12:00-04:00", "type": "Primary product page" } } ], "sourceId": "xai-grok-plans", "sourceValidity": { "status": "stale", "label": "Source retrieval stale", "primary": true, "urlValid": true, "retrievedAt": "2026-07-23T07:12:00-04:00", "ageDays": 88, "authority": "Primary product page", "publisher": "xAI", "published": "Current page" } }, { "key": "promotion|anthropic|claude-sonnet-5-introductory-api-pricing|not-separately-published", "type": "promotion", "provider": "Anthropic", "product": "Claude Sonnet 5 introductory API pricing", "currentTerms": { "offer": "$2 input / $10 output per 1M tokens through August 31, 2026", "currency": "USD", "region": "Not separately published", "eligibility": "Claude API usage under Anthropic’s published terms and regional availability.", "startsAt": null, "endsAt": null, "detail": "Anthropic says standard pricing changes to $3 input and $15 output per million tokens after the introductory period." }, "effectiveFrom": null, "firstObserved": "2026-10-12", "lastObserved": "2026-10-19", "lastChanged": "2026-10-12", "observationCount": 2, "periodCount": 1, "periods": [ { "firstObserved": "2026-10-12", "lastObserved": "2026-10-19", "effectiveFrom": null, "terms": { "offer": "$2 input / $10 output per 1M tokens through August 31, 2026", "currency": "USD", "region": "Not separately published", "eligibility": "Claude API usage under Anthropic’s published terms and regional availability.", "startsAt": null, "endsAt": null, "detail": "Anthropic says standard pricing changes to $3 input and $15 output per million tokens after the introductory period." }, "sourceId": "anthropic-sonnet5", "source": { "id": "anthropic-sonnet5", "title": "Introducing Claude Sonnet 5", "publisher": "Anthropic", "url": "https://www.anthropic.com/news/claude-sonnet-5", "published": "2026-06", "retrieved": "2026-07-23T07:13:00-04:00", "type": "Primary launch and pricing page" } } ], "sourceId": "anthropic-sonnet5", "sourceValidity": { "status": "stale", "label": "Source retrieval stale", "primary": true, "urlValid": true, "retrievedAt": "2026-07-23T07:13:00-04:00", "ageDays": 88, "authority": "Primary launch and pricing page", "publisher": "Anthropic", "published": "2026-06" } }, { "key": "promotion|microsoft|microsoft-365-copilot-business-promotional-pricing|not-separately-published", "type": "promotion", "provider": "Microsoft", "product": "Microsoft 365 Copilot Business promotional pricing", "currentTerms": { "offer": "First-year discount available through September 30, 2026", "currency": "Not separately published", "region": "Not separately published", "eligibility": "Eligible Microsoft 365 business customers; annual commitment and other terms apply.", "startsAt": null, "endsAt": null, "detail": "Microsoft lists promotional pricing for eligible Microsoft 365 Copilot Business plans and add-ons. Discount levels vary by selected plan." }, "effectiveFrom": null, "firstObserved": "2026-10-12", "lastObserved": "2026-10-19", "lastChanged": "2026-10-12", "observationCount": 2, "periodCount": 1, "periods": [ { "firstObserved": "2026-10-12", "lastObserved": "2026-10-19", "effectiveFrom": null, "terms": { "offer": "First-year discount available through September 30, 2026", "currency": "Not separately published", "region": "Not separately published", "eligibility": "Eligible Microsoft 365 business customers; annual commitment and other terms apply.", "startsAt": null, "endsAt": null, "detail": "Microsoft lists promotional pricing for eligible Microsoft 365 Copilot Business plans and add-ons. Discount levels vary by selected plan." }, "sourceId": "microsoft-copilot-business-promo", "source": { "id": "microsoft-copilot-business-promo", "title": "Microsoft 365 Copilot plans and promotional pricing", "publisher": "Microsoft", "url": "https://www.microsoft.com/en-us/microsoft-365-copilot/pricing", "published": "Current promotion through 2026-09-30", "retrieved": "2026-07-23T07:10:00-04:00", "type": "Primary promotion page" } } ], "sourceId": "microsoft-copilot-business-promo", "sourceValidity": { "status": "stale", "label": "Source retrieval stale", "primary": true, "urlValid": true, "retrievedAt": "2026-07-23T07:10:00-04:00", "ageDays": 88, "authority": "Primary promotion page", "publisher": "Microsoft", "published": "Current promotion through 2026-09-30" } } ], "events": [ { "id": "2026-10-19|introduced|api|google-deepmind|gemini-3-8-flash-pro|usd|not-separately-published|standard-published-tier", "editionDate": "2026-10-19", "previousEdition": "2026-10-12", "key": "api|google-deepmind|gemini-3-8-flash-pro|usd|not-separately-published|standard-published-tier", "type": "api", "provider": "Google DeepMind", "product": "Gemini 3.8 (Flash & Pro)", "currency": "USD", "region": "Not separately published", "changeType": "introduced", "fields": [], "previous": null, "current": { "inputPrice": 0.075, "outputPrice": 0.3, "cachedInputPrice": 0.019, "currency": "USD", "unit": "Per 1M tokens", "region": "Not separately published", "tier": "Standard published tier" }, "sourceId": "google-gemini-38-multimodal-release", "effectiveFrom": null, "summary": "Google DeepMind Gemini 3.8 (Flash & Pro) entered the tracked pricing database." } ], "windows": { "7": { "requestedDays": 7, "startDate": "2026-10-13", "endDate": "2026-10-19", "editionCount": 1, "representedDays": 1, "missingDays": 6, "complete": false, "status": "incomplete", "availableFrom": "2026-10-19", "availableTo": "2026-10-19", "counts": { "total": 1, "api": 1, "plan": 0, "promotion": 0, "freeTier": 0, "introduced": 1, "changed": 0, "removed": 0 }, "events": [ { "id": "2026-10-19|introduced|api|google-deepmind|gemini-3-8-flash-pro|usd|not-separately-published|standard-published-tier", "editionDate": "2026-10-19", "previousEdition": "2026-10-12", "key": "api|google-deepmind|gemini-3-8-flash-pro|usd|not-separately-published|standard-published-tier", "type": "api", "provider": "Google DeepMind", "product": "Gemini 3.8 (Flash & Pro)", "currency": "USD", "region": "Not separately published", "changeType": "introduced", "fields": [], "previous": null, "current": { "inputPrice": 0.075, "outputPrice": 0.3, "cachedInputPrice": 0.019, "currency": "USD", "unit": "Per 1M tokens", "region": "Not separately published", "tier": "Standard published tier" }, "sourceId": "google-gemini-38-multimodal-release", "effectiveFrom": null, "summary": "Google DeepMind Gemini 3.8 (Flash & Pro) entered the tracked pricing database." } ], "activeProviders": [ "Google DeepMind" ], "activeCurrencies": [ "USD" ] }, "30": { "requestedDays": 30, "startDate": "2026-09-20", "endDate": "2026-10-19", "editionCount": 2, "representedDays": 8, "missingDays": 28, "complete": false, "status": "incomplete", "availableFrom": "2026-10-12", "availableTo": "2026-10-19", "counts": { "total": 1, "api": 1, "plan": 0, "promotion": 0, "freeTier": 0, "introduced": 1, "changed": 0, "removed": 0 }, "events": [ { "id": "2026-10-19|introduced|api|google-deepmind|gemini-3-8-flash-pro|usd|not-separately-published|standard-published-tier", "editionDate": "2026-10-19", "previousEdition": "2026-10-12", "key": "api|google-deepmind|gemini-3-8-flash-pro|usd|not-separately-published|standard-published-tier", "type": "api", "provider": "Google DeepMind", "product": "Gemini 3.8 (Flash & Pro)", "currency": "USD", "region": "Not separately published", "changeType": "introduced", "fields": [], "previous": null, "current": { "inputPrice": 0.075, "outputPrice": 0.3, "cachedInputPrice": 0.019, "currency": "USD", "unit": "Per 1M tokens", "region": "Not separately published", "tier": "Standard published tier" }, "sourceId": "google-gemini-38-multimodal-release", "effectiveFrom": null, "summary": "Google DeepMind Gemini 3.8 (Flash & Pro) entered the tracked pricing database." } ], "activeProviders": [ "Google DeepMind" ], "activeCurrencies": [ "USD" ] }, "90": { "requestedDays": 90, "startDate": "2026-07-22", "endDate": "2026-10-19", "editionCount": 2, "representedDays": 8, "missingDays": 88, "complete": false, "status": "incomplete", "availableFrom": "2026-10-12", "availableTo": "2026-10-19", "counts": { "total": 1, "api": 1, "plan": 0, "promotion": 0, "freeTier": 0, "introduced": 1, "changed": 0, "removed": 0 }, "events": [ { "id": "2026-10-19|introduced|api|google-deepmind|gemini-3-8-flash-pro|usd|not-separately-published|standard-published-tier", "editionDate": "2026-10-19", "previousEdition": "2026-10-12", "key": "api|google-deepmind|gemini-3-8-flash-pro|usd|not-separately-published|standard-published-tier", "type": "api", "provider": "Google DeepMind", "product": "Gemini 3.8 (Flash & Pro)", "currency": "USD", "region": "Not separately published", "changeType": "introduced", "fields": [], "previous": null, "current": { "inputPrice": 0.075, "outputPrice": 0.3, "cachedInputPrice": 0.019, "currency": "USD", "unit": "Per 1M tokens", "region": "Not separately published", "tier": "Standard published tier" }, "sourceId": "google-gemini-38-multimodal-release", "effectiveFrom": null, "summary": "Google DeepMind Gemini 3.8 (Flash & Pro) entered the tracked pricing database." } ], "activeProviders": [ "Google DeepMind" ], "activeCurrencies": [ "USD" ] } }, "sources": { "openai-gpt56": { "id": "openai-gpt56", "title": "Introducing GPT-5.6", "publisher": "OpenAI", "url": "https://openai.com/index/gpt-5-6/", "published": "2026-07-09", "retrieved": "2026-07-20T19:20:00-04:00", "type": "Primary" }, "openai-chatgpt-pricing": { "id": "openai-chatgpt-pricing", "title": "ChatGPT Plans", "publisher": "OpenAI", "url": "https://openai.com/chatgpt/pricing/", "published": "Current page", "retrieved": "2026-07-20T20:00:00-04:00", "type": "Primary" }, "openai-gpt-red": { "id": "openai-gpt-red", "title": "GPT-Red: Unlocking Self-Improvement for Robustness", "publisher": "OpenAI", "url": "https://openai.com/", "published": "2026-07-15", "retrieved": "2026-07-20T19:45:00-04:00", "type": "Primary" }, "anthropic-fable5": { "id": "anthropic-fable5", "title": "Claude Fable 5 and Mythos 5", "publisher": "Anthropic", "url": "https://www.anthropic.com/news/claude-fable-5-mythos-5", "published": "2026-06", "retrieved": "2026-07-20T19:24:00-04:00", "type": "Primary" }, "anthropic-claude": { "id": "anthropic-claude", "title": "Claude product overview", "publisher": "Anthropic", "url": "https://claude.com/", "published": "Current page", "retrieved": "2026-07-20T20:00:00-04:00", "type": "Primary" }, "xai-grok45": { "id": "xai-grok45", "title": "Grok 4.5", "publisher": "xAI", "url": "https://x.ai/news/grok-4-5", "published": "2026-07-16", "retrieved": "2026-07-20T19:28:00-04:00", "type": "Primary" }, "xai-pricing": { "id": "xai-pricing", "title": "xAI API pricing", "publisher": "xAI", "url": "https://docs.x.ai/developers/pricing", "published": "2026-07", "retrieved": "2026-07-20T20:00:00-04:00", "type": "Primary" }, "xai-grok-build": { "id": "xai-grok-build", "title": "Grok Build release", "publisher": "xAI", "url": "https://x.ai/news", "published": "2026-07-15", "retrieved": "2026-07-20T19:30:00-04:00", "type": "Primary" }, "google-gemini35": { "id": "google-gemini35", "title": "Gemini 3.5", "publisher": "Google", "url": "https://blog.google/innovation-and-ai/models-and-research/gemini-models/gemini-3-5/", "published": "2026-05-19", "retrieved": "2026-07-20T19:31:00-04:00", "type": "Primary" }, "google-ai-plans": { "id": "google-ai-plans", "title": "Google AI plans", "publisher": "Google", "url": "https://one.google.com/about/google-ai-plans/", "published": "Current page", "retrieved": "2026-07-20T20:00:00-04:00", "type": "Primary" }, "google-parallel-search": { "id": "google-parallel-search", "title": "Parallel web search grounding update", "publisher": "Google", "url": "https://blog.google/technology/ai/", "published": "2026-07", "retrieved": "2026-07-20T19:48:00-04:00", "type": "Primary" }, "deepseek-v4": { "id": "deepseek-v4", "title": "DeepSeek V4 Preview", "publisher": "DeepSeek", "url": "https://api-docs.deepseek.com/news/news260424/", "published": "2026-04-24", "retrieved": "2026-07-20T19:35:00-04:00", "type": "Primary" }, "deepseek-models": { "id": "deepseek-models", "title": "DeepSeek API model and alias documentation", "publisher": "DeepSeek", "url": "https://api-docs.deepseek.com/quick_start/pricing/", "published": "Current page", "retrieved": "2026-07-20T19:36:00-04:00", "type": "Primary" }, "amd-microsoft-helios": { "id": "amd-microsoft-helios", "title": "Microsoft to deploy AMD Helios Rackscale Solution on Azure", "publisher": "AMD", "url": "https://www.amd.com/en/newsroom/press-releases.html", "published": "2026-07-20", "retrieved": "2026-07-20T19:40:00-04:00", "type": "Primary" }, "reuters-waic-governance": { "id": "reuters-waic-governance", "title": "China positions itself in global AI governance at WAIC", "publisher": "Reuters", "url": "https://www.reuters.com/technology/artificial-intelligence/", "published": "2026-07-17", "retrieved": "2026-07-20T19:43:00-04:00", "type": "Secondary" }, "reuters-huawei-waic": { "id": "reuters-huawei-waic", "title": "Huawei presents Atlas 950 SuperPoD at WAIC", "publisher": "Reuters", "url": "https://www.reuters.com/technology/artificial-intelligence/", "published": "2026-07-16", "retrieved": "2026-07-20T19:44:00-04:00", "type": "Secondary" }, "reuters-earnings-watch": { "id": "reuters-earnings-watch", "title": "Alphabet and Intel earnings put AI trade to the test", "publisher": "Reuters", "url": "https://www.reuters.com/technology/artificial-intelligence/", "published": "2026-07-17", "retrieved": "2026-07-20T19:44:00-04:00", "type": "Secondary" }, "nvidia-rtx50": { "id": "nvidia-rtx50", "title": "GeForce RTX 50 Series announcement", "publisher": "NVIDIA", "url": "https://www.nvidia.com/en-us/geforce/graphics-cards/50-series/", "published": "Current product page", "retrieved": "2026-07-20T19:50:00-04:00", "type": "Primary" }, "nvidia-rtx5070": { "id": "nvidia-rtx5070", "title": "GeForce RTX 5070", "publisher": "NVIDIA", "url": "https://www.nvidia.com/en-us/geforce/graphics-cards/50-series/rtx-5070-family/", "published": "Current product page", "retrieved": "2026-07-20T19:50:00-04:00", "type": "Primary" }, "nvidia-rtx5050": { "id": "nvidia-rtx5050", "title": "GeForce RTX 5050 announcement", "publisher": "NVIDIA", "url": "https://www.nvidia.com/en-us/geforce/news/", "published": "2026-07", "retrieved": "2026-07-20T19:51:00-04:00", "type": "Primary" }, "amd-halo": { "id": "amd-halo", "title": "Ryzen AI Halo Developer Platform", "publisher": "AMD / Micro Center", "url": "https://www.amd.com/en/developer/resources/ryzen-ai-software.html", "published": "2026", "retrieved": "2026-07-20T19:52:00-04:00", "type": "Primary / retailer" }, "apple-m5pro": { "id": "apple-m5pro", "title": "MacBook Pro with M5 Pro and M5 Max", "publisher": "Apple", "url": "https://www.apple.com/macbook-pro/", "published": "2026", "retrieved": "2026-07-20T19:54:00-04:00", "type": "Primary" }, "apple-neo": { "id": "apple-neo", "title": "MacBook Neo", "publisher": "Apple", "url": "https://www.apple.com/macbook-neo/", "published": "2026", "retrieved": "2026-07-20T19:54:00-04:00", "type": "Primary" }, "market-feed": { "id": "market-feed", "title": "Latest available US market quote feed", "publisher": "Market data feed", "url": "https://www.nasdaq.com/market-activity", "published": "2026-07-20", "retrieved": "2026-07-20T19:59:00-04:00", "type": "Market data" }, "alibaba-qwen37": { "id": "alibaba-qwen37", "title": "Alibaba Cloud Model Studio — Qwen flagship models", "publisher": "Alibaba Cloud", "url": "https://help.aliyun.com/en/model-studio/what-is-model-studio", "published": "Updated 2026-07-10", "retrieved": "2026-07-20T20:00:00-04:00", "type": "Primary" }, "moonshot-kimi-k3": { "id": "moonshot-kimi-k3", "title": "Kimi API platform — Kimi K3 model and pricing", "publisher": "Moonshot AI", "url": "https://platform.moonshot.cn/", "published": "Current page", "retrieved": "2026-07-20T20:00:00-04:00", "type": "Primary" }, "zhipu-glm52": { "id": "zhipu-glm52", "title": "Zhipu BigModel — GLM-5.2 model overview", "publisher": "Zhipu AI", "url": "https://docs.bigmodel.cn/cn/guide/start/model-overview", "published": "Current page", "retrieved": "2026-07-20T20:00:00-04:00", "type": "Primary" }, "zhipu-glm-ocr": { "id": "zhipu-glm-ocr", "title": "Zhipu BigModel — GLM-OCR", "publisher": "Zhipu AI", "url": "https://docs.bigmodel.cn/cn/guide/models/vlm/glm-ocr", "published": "Current page", "retrieved": "2026-07-20T20:00:00-04:00", "type": "Primary" }, "baidu-ernie50": { "id": "baidu-ernie50", "title": "Baidu Qianfan — ERNIE 5.0 model list and international price", "publisher": "Baidu AI Cloud", "url": "https://intl.cloud.baidu.com/en/doc/qianfan/s/7m95lyy43-intl-en", "published": "Current page", "retrieved": "2026-07-20T20:00:00-04:00", "type": "Primary" }, "bytedance-doubao-seed21": { "id": "bytedance-doubao-seed21", "title": "Volcengine Ark — Doubao Seed 2.1 model catalog", "publisher": "ByteDance / Volcengine", "url": "https://www.volcengine.com/docs/82379/1330310", "published": "Current page", "retrieved": "2026-07-20T20:00:00-04:00", "type": "Primary" }, "minimax-m3": { "id": "minimax-m3", "title": "MiniMax API — MiniMax-M3 model release and catalog", "publisher": "MiniMax", "url": "https://platform.minimax.io/docs/release-notes/models", "published": "2026-06-01", "retrieved": "2026-07-20T20:00:00-04:00", "type": "Primary" }, "stepfun-37-flash": { "id": "stepfun-37-flash", "title": "StepFun — Step 3.7 Flash pricing and model documentation", "publisher": "StepFun", "url": "https://platform.stepfun.com/docs/zh/guides/pricing/details", "published": "Current page", "retrieved": "2026-07-20T20:00:00-04:00", "type": "Primary" }, "tencent-hunyuan-a13b": { "id": "tencent-hunyuan-a13b", "title": "Tencent Cloud — Hunyuan A13B model overview", "publisher": "Tencent Cloud", "url": "https://cloud.tencent.com/document/product/1729/104753", "published": "Current page", "retrieved": "2026-07-20T20:00:00-04:00", "type": "Primary" }, "openai-hf-incident": { "id": "openai-hf-incident", "title": "OpenAI and Hugging Face partner to address security incident during model evaluation", "publisher": "OpenAI", "url": "https://openai.com/index/hugging-face-model-evaluation-security-incident/", "published": "2026-07-21", "retrieved": "2026-07-21T19:35:00-04:00", "type": "Primary" }, "reuters-china-controls": { "id": "reuters-china-controls", "title": "China considers tighter export controls on AI models and chips, FT reports", "publisher": "Reuters", "url": "https://www.reuters.com/world/asia-pacific/china-considers-tighter-export-controls-ai-models-chips-ft-reports-2026-07-21/", "published": "2026-07-21", "retrieved": "2026-07-21T19:40:00-04:00", "type": "Secondary reporting" }, "reuters-tsmc-prices": { "id": "reuters-tsmc-prices", "title": "TSMC to raise chipmaking prices by up to 10% in 2027, Nikkei Asia reports", "publisher": "Reuters", "url": "https://www.reuters.com/world/asia-pacific/tsmc-raise-chipmaking-prices-by-up-10-2027-nikkei-asia-reports-2026-07-21/", "published": "2026-07-21", "retrieved": "2026-07-21T19:42:00-04:00", "type": "Secondary reporting" }, "supermicro-q4-update": { "id": "supermicro-q4-update", "title": "Supermicro Provides Fourth Quarter of Fiscal Year 2026 Preliminary Business Update", "publisher": "Supermicro Investor Relations", "url": "https://ir.supermicro.com/news/news-details/2026/Supermicro-Provides-Fourth-Quarter-of-Fiscal-Year-2026-Preliminary-Business-Update/default.aspx", "published": "2026-07-21", "retrieved": "2026-07-21T19:44:00-04:00", "type": "Primary" }, "reuters-alphabet-earnings": { "id": "reuters-alphabet-earnings", "title": "Alphabet's Gemini delay, spending worries loom over earnings", "publisher": "Reuters", "url": "https://www.reuters.com/business/alphabets-gemini-delay-spending-worries-loom-over-earnings-2026-07-21/", "published": "2026-07-21", "retrieved": "2026-07-21T19:47:00-04:00", "type": "Secondary reporting" }, "reuters-anthropic-patent": { "id": "reuters-anthropic-patent", "title": "Anthropic sued for infringing neural network technology patents", "publisher": "Reuters", "url": "https://www.reuters.com/legal/government/university-tennessee-sues-anthropic-over-neural-network-technology-2026-07-21/", "published": "2026-07-21", "retrieved": "2026-07-21T19:50:00-04:00", "type": "Secondary reporting" }, "reuters-humanoid-funding": { "id": "reuters-humanoid-funding", "title": "Robotics startup Humanoid raises $152 million Series A round at $1.35 billion valuation", "publisher": "Reuters", "url": "https://www.reuters.com/business/robotics-startup-humanoid-raises-152-million-series-round-135-billion-valuation-2026-07-21/", "published": "2026-07-21", "retrieved": "2026-07-21T19:52:00-04:00", "type": "Secondary reporting" }, "market-msft": { "id": "market-msft", "title": "Microsoft stock quote", "publisher": "Nasdaq", "url": "https://www.nasdaq.com/market-activity/stocks/msft", "published": "Live market page", "retrieved": "2026-07-21T19:48:00-04:00", "type": "Market data" }, "market-googl": { "id": "market-googl", "title": "Alphabet Class A stock quote", "publisher": "Nasdaq", "url": "https://www.nasdaq.com/market-activity/stocks/googl", "published": "Live market page", "retrieved": "2026-07-21T19:48:00-04:00", "type": "Market data" }, "market-meta": { "id": "market-meta", "title": "Meta Platforms stock quote", "publisher": "Nasdaq", "url": "https://www.nasdaq.com/market-activity/stocks/meta", "published": "Live market page", "retrieved": "2026-07-21T19:48:00-04:00", "type": "Market data" }, "market-amzn": { "id": "market-amzn", "title": "Amazon stock quote", "publisher": "Nasdaq", "url": "https://www.nasdaq.com/market-activity/stocks/amzn", "published": "Live market page", "retrieved": "2026-07-21T19:48:00-04:00", "type": "Market data" }, "market-tsm": { "id": "market-tsm", "title": "TSMC ADR stock quote", "publisher": "NYSE", "url": "https://www.nyse.com/quote/XNYS:TSM", "published": "Live market page", "retrieved": "2026-07-21T19:48:00-04:00", "type": "Market data" }, "market-orcl": { "id": "market-orcl", "title": "Oracle stock quote", "publisher": "NYSE", "url": "https://www.nyse.com/quote/XNYS:ORCL", "published": "Live market page", "retrieved": "2026-07-21T19:48:00-04:00", "type": "Market data" }, "openai-presence": { "id": "openai-presence", "title": "Introducing OpenAI Presence", "publisher": "OpenAI", "url": "https://openai.com/index/introducing-openai-presence/", "published": "2026-07-22", "retrieved": "2026-07-23T06:35:00-04:00", "type": "Primary" }, "openai-national-science": { "id": "openai-national-science", "title": "Advancing the next era of national science", "publisher": "OpenAI", "url": "https://openai.com/index/advancing-the-next-era-of-national-science/", "published": "2026-07-22", "retrieved": "2026-07-23T06:38:00-04:00", "type": "Primary" }, "amd-anthropic": { "id": "amd-anthropic", "title": "AMD and Anthropic announce strategic partnership for up to 2 GW of MI450 Series GPUs", "publisher": "AMD Newsroom", "url": "https://newsroom.amd.com/news/amd-anthropic-strategic-partnership/", "published": "2026-07-22", "retrieved": "2026-07-23T06:42:00-04:00", "type": "Primary" }, "google-q2-2026": { "id": "google-q2-2026", "title": "Alphabet Q2 2026 earnings call — CEO remarks", "publisher": "Google / Alphabet", "url": "https://blog.google/company-news/inside-google/message-ceo/alphabet-earnings-q2-2026/", "published": "2026-07-22", "retrieved": "2026-07-23T06:45:00-04:00", "type": "Primary" }, "qwen-image3": { "id": "qwen-image3", "title": "Qwen-Image-3.0: Rich Content, Authentic Details, Deep Knowledge", "publisher": "Alibaba Cloud / Qwen", "url": "https://www.alibabacloud.com/blog/qwen-image-3-0-rich-content-authentic-details-deep-knowledge_603385", "published": "2026-07-22", "retrieved": "2026-07-23T06:50:00-04:00", "type": "Primary provider publication" }, "qwen-audio3": { "id": "qwen-audio3", "title": "Qwen-Audio-3.0-TTS: More Multilingual, Easier to Direct", "publisher": "Alibaba Cloud / Qwen", "url": "https://www.alibabacloud.com/blog/qwen-audio-3-0-tts-more-multilingual-easier-to-direct_603379", "published": "2026-07-21", "retrieved": "2026-07-23T06:52:00-04:00", "type": "Primary provider publication" }, "reuters-server-cpu-china": { "id": "reuters-server-cpu-china", "title": "Intel and AMD sign longer-term server CPU deals with Chinese clients as prices surge", "publisher": "Reuters", "url": "https://www.reuters.com/legal/transactional/intel-amd-sign-long-term-server-cpu-deals-with-chinese-clients-prices-surge-2026-07-23/", "published": "2026-07-23", "retrieved": "2026-07-23T07:02:00-04:00", "type": "Secondary reporting" }, "ap-markets-jul22": { "id": "ap-markets-jul22", "title": "How major US stock indexes fared Wednesday, July 22, 2026", "publisher": "Associated Press", "url": "https://apnews.com/article/87e0cd1ef1994f0fcbfb8f025064f64f", "published": "2026-07-22", "retrieved": "2026-07-23T07:05:00-04:00", "type": "Market summary" }, "market-nvda": { "id": "market-nvda", "title": "NVIDIA July 22, 2026 closing quote", "publisher": "Yahoo Finance", "url": "https://finance.yahoo.com/quote/NVDA/", "published": "2026-07-22 close", "retrieved": "2026-07-23T07:05:00-04:00", "type": "Market data" }, "market-amd": { "id": "market-amd", "title": "AMD July 22, 2026 closing quote", "publisher": "ChartExchange", "url": "https://chartexchange.com/symbol/nasdaq-amd/", "published": "2026-07-22 close", "retrieved": "2026-07-23T07:06:00-04:00", "type": "Market data" }, "market-smci": { "id": "market-smci", "title": "Super Micro Computer July 22, 2026 closing quote", "publisher": "ChartExchange", "url": "https://chartexchange.com/symbol/nasdaq-smci/historical/", "published": "2026-07-22 close", "retrieved": "2026-07-23T07:06:00-04:00", "type": "Market data" }, "market-avgo": { "id": "market-avgo", "title": "Broadcom July 22, 2026 closing quote", "publisher": "MarketWatch", "url": "https://www.marketwatch.com/investing/stock/avgo", "published": "2026-07-22 close", "retrieved": "2026-07-23T07:07:00-04:00", "type": "Market data" }, "microsoft-copilot-pro": { "id": "microsoft-copilot-pro", "title": "Microsoft Copilot Pro plan and pricing", "publisher": "Microsoft", "url": "https://www.microsoft.com/en-us/store/b/copilotpro", "published": "Current page", "retrieved": "2026-07-23T07:09:00-04:00", "type": "Primary plan page" }, "microsoft-copilot-business-promo": { "id": "microsoft-copilot-business-promo", "title": "Microsoft 365 Copilot plans and promotional pricing", "publisher": "Microsoft", "url": "https://www.microsoft.com/en-us/microsoft-365-copilot/pricing", "published": "Current promotion through 2026-09-30", "retrieved": "2026-07-23T07:10:00-04:00", "type": "Primary promotion page" }, "perplexity-plans": { "id": "perplexity-plans", "title": "Which Perplexity subscription plan is right for you?", "publisher": "Perplexity Support", "url": "https://www.perplexity.ai/help-center/en/articles/11187416-which-perplexity-subscription-plan-is-right-for-you", "published": "Updated 2026-07-22", "retrieved": "2026-07-23T07:11:00-04:00", "type": "Primary plan documentation" }, "xai-grok-plans": { "id": "xai-grok-plans", "title": "Grok product and plan page", "publisher": "xAI", "url": "https://grok.com/", "published": "Current page", "retrieved": "2026-07-23T07:12:00-04:00", "type": "Primary product page" }, "anthropic-sonnet5": { "id": "anthropic-sonnet5", "title": "Introducing Claude Sonnet 5", "publisher": "Anthropic", "url": "https://www.anthropic.com/news/claude-sonnet-5", "published": "2026-06", "retrieved": "2026-07-23T07:13:00-04:00", "type": "Primary launch and pricing page" }, "openai-health-chatgpt": { "id": "openai-health-chatgpt", "title": "Launching Health in ChatGPT", "publisher": "OpenAI", "url": "https://openai.com/index/health-in-chatgpt/", "published": "2026-07-23", "retrieved": "2026-07-23T19:55:00-04:00", "type": "Primary" }, "google-atlas-ai-economy": { "id": "google-atlas-ai-economy", "title": "Understanding the AI economy — the first ATLAS report", "publisher": "Google", "url": "https://blog.google/innovation-and-ai/technology/research/understanding-the-ai-economy/", "published": "2026-07-23", "retrieved": "2026-07-23T20:00:00-04:00", "type": "Primary" }, "deepseek-v4-pricing-retirement": { "id": "deepseek-v4-pricing-retirement", "title": "DeepSeek V4 models, pricing and legacy alias retirement", "publisher": "DeepSeek API Docs", "url": "https://api-docs.deepseek.com/quick_start/pricing", "published": "Current documentation; alias retirement 2026-07-24 15:59 UTC", "retrieved": "2026-07-23T20:05:00-04:00", "type": "Primary" }, "ftc-ai-accuracy-policy": { "id": "ftc-ai-accuracy-policy", "title": "FTC seeks public comment on policy statement addressing AI accuracy", "publisher": "Federal Trade Commission", "url": "https://www.ftc.gov/news-events/news/press-releases/2026/07/ftc-seeks-public-comment-policy-statement-addressing-ai-accuracy", "published": "2026-07; comments due 2026-07-31", "retrieved": "2026-07-23T20:08:00-04:00", "type": "Government source" }, "ap-markets-jul23": { "id": "ap-markets-jul23", "title": "How major US stock indexes fared Thursday, July 23, 2026", "publisher": "Associated Press", "url": "https://apnews.com/article/wall-street-dow-nasdaq-stocks-3cb4bc150707e0a4378312fbb825f5e4", "published": "2026-07-23", "retrieved": "2026-07-23T20:12:00-04:00", "type": "Market summary" }, "market-googl-jul23": { "id": "market-googl-jul23", "title": "Alphabet July 23, 2026 closing quote", "publisher": "Nasdaq", "url": "https://www.nasdaq.com/market-activity/stocks/googl/historical", "published": "2026-07-23 close", "retrieved": "2026-07-23T20:15:00-04:00", "type": "Historical market data" }, "market-nvda-jul23": { "id": "market-nvda-jul23", "title": "NVIDIA July 23, 2026 closing quote", "publisher": "Nasdaq", "url": "https://www.nasdaq.com/market-activity/stocks/nvda/historical", "published": "2026-07-23 close", "retrieved": "2026-07-23T20:16:00-04:00", "type": "Historical market data" }, "market-amd-jul23": { "id": "market-amd-jul23", "title": "AMD July 23, 2026 closing quote", "publisher": "Nasdaq", "url": "https://www.nasdaq.com/market-activity/stocks/amd/historical", "published": "2026-07-23 close", "retrieved": "2026-07-23T20:17:00-04:00", "type": "Historical market data" }, "market-smci-jul23": { "id": "market-smci-jul23", "title": "Super Micro Computer July 23, 2026 closing quote", "publisher": "Nasdaq", "url": "https://www.nasdaq.com/market-activity/stocks/smci/historical", "published": "2026-07-23 close", "retrieved": "2026-07-23T20:18:00-04:00", "type": "Historical market data" }, "market-avgo-jul23": { "id": "market-avgo-jul23", "title": "Broadcom July 23, 2026 closing quote", "publisher": "Nasdaq", "url": "https://www.nasdaq.com/market-activity/stocks/avgo/historical", "published": "2026-07-23 close", "retrieved": "2026-07-23T20:19:00-04:00", "type": "Historical market data" }, "market-msft-jul23": { "id": "market-msft-jul23", "title": "Microsoft July 23, 2026 closing quote", "publisher": "Nasdaq", "url": "https://www.nasdaq.com/market-activity/stocks/msft/historical", "published": "2026-07-23 close", "retrieved": "2026-07-23T20:20:00-04:00", "type": "Historical market data" }, "meta-ai-actions-jul24": { "id": "meta-ai-actions-jul24", "title": "Meta AI Doesn’t Just Think, It Acts", "publisher": "Meta", "url": "https://about.fb.com/news/2026/07/meta-ai-muse-spark-doesnt-just-think-it-acts/", "published": "2026-07-24", "retrieved": "2026-07-25T03:25:00-04:00", "type": "Primary" }, "google-eu-ai-transparency-code": { "id": "google-eu-ai-transparency-code", "title": "Google signs the EU AI Act Code of Practice on Transparency of AI-Generated Content", "publisher": "Google", "url": "https://blog.google/company-news/outreach-and-initiatives/public-policy/eu-ai-act-transparency-code-of-practice/", "published": "2026-07-24", "retrieved": "2026-07-25T03:30:00-04:00", "type": "Primary" }, "amd-advancing-ai-2026": { "id": "amd-advancing-ai-2026", "title": "AMD Advancing AI 2026", "publisher": "AMD", "url": "https://www.amd.com/en/corporate/events/advancing-ai.html", "published": "2026-07-22 to 2026-07-23", "retrieved": "2026-07-25T03:35:00-04:00", "type": "Primary" }, "deepseek-v4-post-retirement": { "id": "deepseek-v4-post-retirement", "title": "DeepSeek V4 models, pricing and legacy alias retirement", "publisher": "DeepSeek API Docs", "url": "https://api-docs.deepseek.com/quick_start/pricing", "published": "Current documentation; legacy aliases retired 2026-07-24 15:59 UTC", "retrieved": "2026-07-25T03:40:00-04:00", "type": "Primary" }, "market-summary-jul24": { "id": "market-summary-jul24", "title": "Broadcom shares fall as the S&P 500 and Dow finish slightly higher", "publisher": "MarketWatch", "url": "https://www.marketwatch.com/data-news/broadcom-inc-stock-underperforms-friday-when-compared-to-competitors-9e8602a4-a84725bbedab", "published": "2026-07-24 close", "retrieved": "2026-07-25T03:45:00-04:00", "type": "Market summary" }, "market-googl-jul24": { "id": "market-googl-jul24", "title": "Alphabet July 24, 2026 historical quote", "publisher": "Nasdaq", "url": "https://www.nasdaq.com/market-activity/stocks/googl/historical", "published": "2026-07-24 close", "retrieved": "2026-07-25T03:46:00-04:00", "type": "Historical market data" }, "market-nvda-jul24": { "id": "market-nvda-jul24", "title": "NVIDIA July 24, 2026 historical quote", "publisher": "Nasdaq", "url": "https://www.nasdaq.com/market-activity/stocks/nvda/historical", "published": "2026-07-24 close", "retrieved": "2026-07-25T03:47:00-04:00", "type": "Historical market data" }, "market-amd-jul24": { "id": "market-amd-jul24", "title": "AMD July 24, 2026 historical quote", "publisher": "Nasdaq", "url": "https://www.nasdaq.com/market-activity/stocks/amd/historical", "published": "2026-07-24 close", "retrieved": "2026-07-25T03:48:00-04:00", "type": "Historical market data" }, "market-smci-jul24": { "id": "market-smci-jul24", "title": "Super Micro Computer July 24, 2026 historical quote", "publisher": "Nasdaq", "url": "https://www.nasdaq.com/market-activity/stocks/smci/historical", "published": "2026-07-24 close", "retrieved": "2026-07-25T03:49:00-04:00", "type": "Historical market data" }, "market-avgo-jul24": { "id": "market-avgo-jul24", "title": "Broadcom July 24, 2026 historical quote", "publisher": "Nasdaq", "url": "https://www.nasdaq.com/market-activity/stocks/avgo/historical", "published": "2026-07-24 close", "retrieved": "2026-07-25T03:50:00-04:00", "type": "Historical market data" }, "nvidia-sk-ai-factory-jul24": { "id": "nvidia-sk-ai-factory-jul24", "title": "SK Group and NVIDIA Expand Strategic Partnership Across AI Factories and Next-Generation Memory", "publisher": "NVIDIA Newsroom", "url": "https://nvidianews.nvidia.com/news/sk-group-and-nvidia-expand-strategic-partnership-across-ai-factories-and-next-generation-memory", "published": "2026-07-24", "retrieved": "2026-07-26T05:20:00-04:00", "type": "Primary" }, "nvidia-korea-summit-jul24": { "id": "nvidia-korea-summit-jul24", "title": "At AI Summit, South Korea Outlines Its AI Future With NVIDIA and Partners", "publisher": "NVIDIA Blog", "url": "https://blogs.nvidia.com/blog/ai-summit-korea-partners-and-nvidia/", "published": "2026-07-23; updated 2026-07-24", "retrieved": "2026-07-26T05:30:00-04:00", "type": "Primary" }, "nvidia-kaist-lab-jul23": { "id": "nvidia-kaist-lab-jul23", "title": "NVIDIA and KAIST Launch Joint AI Research Lab to Accelerate AI Innovation in Korea", "publisher": "NVIDIA Newsroom", "url": "https://nvidianews.nvidia.com/news/nvidia-and-kaist-launch-joint-ai-research-lab-to-accelerate-ai-innovation-in-korea", "published": "2026-07-23", "retrieved": "2026-07-26T05:35:00-04:00", "type": "Primary" }, "wsj-openai-ohio-backstop-jul26": { "id": "wsj-openai-ohio-backstop-jul26", "title": "Nvidia in Talks With OpenAI to Guarantee $250 Billion Financing for Data Center", "publisher": "The Wall Street Journal", "url": "https://www.wsj.com/tech/ai/nvidia-in-talks-with-openai-to-guarantee-250-billion-financing-for-data-center-3dd6eae3", "published": "2026-07-26", "retrieved": "2026-07-26T20:30:00-04:00", "type": "Secondary reporting" }, "reuters-openai-ohio-backstop-jul26": { "id": "reuters-openai-ohio-backstop-jul26", "title": "Nvidia in talks with OpenAI to guarantee $250 billion financing for data center, WSJ reports", "publisher": "Reuters", "url": "https://wtaq.com/2026/07/26/nvidia-in-talks-with-openai-to-guarantee-250-billion-financing-for-data-center-wsj-reports/", "published": "2026-07-26", "retrieved": "2026-07-26T20:35:00-04:00", "type": "Secondary reporting" }, "theinformation-openai-ohio-campus-jun09": { "id": "theinformation-openai-ohio-campus-jun09", "title": "OpenAI in Talks to Lease 10 Gigawatt Ohio Data Center with Backing From Nvidia", "publisher": "The Information", "url": "https://www.theinformation.com/articles/openai-talks-lease-10-gigawatt-ohio-data-center-backing-nvidia/", "published": "2026-06-09", "retrieved": "2026-07-26T20:38:00-04:00", "type": "Secondary reporting" }, "nvidia-open-secure-ai-alliance-jul27": { "id": "nvidia-open-secure-ai-alliance-jul27", "title": "Industry Leaders Unite in Open Secure AI Alliance for AI Safety and Security", "publisher": "NVIDIA Blog", "url": "https://blogs.nvidia.com/blog/open-secure-ai-alliance/", "published": "2026-07-27", "retrieved": "2026-07-27T19:45:00-04:00", "type": "Primary" }, "nvidia-ssi-partnership-jul27": { "id": "nvidia-ssi-partnership-jul27", "title": "Ilya Sutskever’s Safe Superintelligence Inc. and NVIDIA Announce Long-Term Strategic Partnership", "publisher": "NVIDIA Newsroom", "url": "https://nvidianews.nvidia.com/news/ilya-sutskevers-safe-superintelligence-inc-and-nvidia-announce-long-term-strategic-partnership", "published": "2026-07-27", "retrieved": "2026-07-27T19:55:00-04:00", "type": "Primary" }, "openai-work-frontier-jul27": { "id": "openai-work-frontier-jul27", "title": "How AI is expanding what people do at work", "publisher": "OpenAI", "url": "https://openai.com/index/how-ai-is-expanding-what-people-do-at-work/", "published": "2026-07-27", "retrieved": "2026-07-27T20:05:00-04:00", "type": "Primary" }, "reuters-altman-huang-warner-jul27": { "id": "reuters-altman-huang-warner-jul27", "title": "OpenAI CEO Sam Altman and NVIDIA CEO Jensen Huang to meet Senator Mark Warner", "publisher": "Reuters", "url": "https://www.reuters.com/business/openais-sam-altman-meet-with-senate-intelligence-committees-top-democrat-2026-07-27/", "published": "2026-07-27", "retrieved": "2026-07-27T20:15:00-04:00", "type": "Secondary reporting" }, "ap-us-market-close-jul27": { "id": "ap-us-market-close-jul27", "title": "How major US stock indexes fared Monday 7/27/2026", "publisher": "The Associated Press", "url": "https://apnews.com/article/wall-street-stocks-dow-nasdaq-e7325919557efc52535a96f09ab38b0d", "published": "2026-07-27 close", "retrieved": "2026-07-27T20:25:00-04:00", "type": "Market summary" }, "market-nvda-jul27": { "id": "market-nvda-jul27", "title": "NVIDIA July 27, 2026 historical quote", "publisher": "Investing.com", "url": "https://www.investing.com/equities/nvidia-corp-historical-data?symbol=NVDA", "published": "2026-07-27 close", "retrieved": "2026-07-27T20:30:00-04:00", "type": "Historical market data" }, "meta-blackrock-el-paso-jul28": { "id": "meta-blackrock-el-paso-jul28", "title": "Meta Announces New Strategic Venture with BlackRock to Develop Data Center in El Paso", "publisher": "Meta Investor Relations", "url": "https://investor.atmeta.com/investor-news/press-release-details/2026/Meta-Announces-New-Strategic-Venture-with-BlackRock-to-Develop-Data-Center-in-El-Paso/default.aspx", "published": "2026-07-28", "retrieved": "2026-07-29T00:02:00-04:00", "type": "Primary" }, "pacing-frontier-jul28": { "id": "pacing-frontier-jul28", "title": "Pacing the Frontier", "publisher": "Pacing the Frontier", "url": "https://www.pacingthefrontier.com/", "published": "2026-07-28", "retrieved": "2026-07-29T00:05:00-04:00", "type": "Employee statement" }, "mcp-releases-status-jul28": { "id": "mcp-releases-status-jul28", "title": "Releases · modelcontextprotocol/modelcontextprotocol", "publisher": "Model Context Protocol on GitHub", "url": "https://github.com/modelcontextprotocol/modelcontextprotocol/releases", "published": "Status at 2026-07-29 cutoff", "retrieved": "2026-07-29T00:08:00-04:00", "type": "Official repository" }, "mcp-versioning-status-jul28": { "id": "mcp-versioning-status-jul28", "title": "Versioning - Model Context Protocol", "publisher": "Model Context Protocol", "url": "https://modelcontextprotocol.io/docs/learn/versioning", "published": "Status at 2026-07-29 cutoff", "retrieved": "2026-07-29T00:09:00-04:00", "type": "Official documentation" }, "fcc-robots-inverters-jul28": { "id": "fcc-robots-inverters-jul28", "title": "FCC bans humanoid robots from China, citing unacceptable risks", "publisher": "The Guardian", "url": "https://www.theguardian.com/us-news/2026/jul/28/fcc-ban-humanoid-robots-china", "published": "2026-07-28", "retrieved": "2026-07-29T00:11:00-04:00", "type": "Secondary reporting with FCC determination link" }, "reuters-chip-rout-jul28": { "id": "reuters-chip-rout-jul28", "title": "Chip rout", "publisher": "Reuters", "url": "https://www.reuters.com/podcasts/reuters-morning-bid/chip-rout-2026-07-28/", "published": "2026-07-28", "retrieved": "2026-07-29T00:13:00-04:00", "type": "Market analysis" }, "ap-us-market-close-jul28": { "id": "ap-us-market-close-jul28", "title": "How major US stock indexes fared Tuesday 7/28/2026", "publisher": "The Associated Press", "url": "https://apnews.com/article/wall-street-stocks-dow-nasdaq-d2a114bce818cab6ee9583f4815cb89a", "published": "2026-07-28 close", "retrieved": "2026-07-29T00:15:00-04:00", "type": "Market summary" }, "market-qcom-jul28": { "id": "market-qcom-jul28", "title": "Qualcomm Inc. stock underperforms Tuesday when compared to competitors", "publisher": "MarketWatch", "url": "https://www.marketwatch.com/data-news/qualcomm-inc-stock-underperforms-tuesday-when-compared-to-competitors-55f7faa8-4edc212b9770", "published": "2026-07-28 close", "retrieved": "2026-07-29T00:17:00-04:00", "type": "Historical market data" }, "reuters-wall-street-fed-ai-jul29": { "id": "reuters-wall-street-fed-ai-jul29", "title": "Wall Street closes down sharply after Fed holds rates unchanged", "publisher": "Reuters", "url": "https://www.reuters.com/business/sp-500-nasdaq-futures-inch-up-before-fed-decision-chip-stocks-wobble-2026-07-29/", "published": "2026-07-29", "retrieved": "2026-07-29T19:10:00-04:00", "type": "Market and earnings reporting" }, "ap-us-market-close-jul29": { "id": "ap-us-market-close-jul29", "title": "How major US stock indexes fared Wednesday 7/29/2026", "publisher": "The Associated Press", "url": "https://apnews.com/article/a483060a0aa9c24c50f589b0b7713497", "published": "2026-07-29 close", "retrieved": "2026-07-29T19:12:00-04:00", "type": "Market summary" }, "meta-q2-event-jul29": { "id": "meta-q2-event-jul29", "title": "Q2 2026 Earnings Call", "publisher": "Meta Investor Relations", "url": "https://investor.atmeta.com/investor-events/event-details/2026/Q2-2026-Earnings-Call/default.aspx", "published": "2026-07-29", "retrieved": "2026-07-29T19:15:00-04:00", "type": "Primary" }, "microsoft-fy26-q4-event-jul29": { "id": "microsoft-fy26-q4-event-jul29", "title": "Microsoft announces quarterly earnings release date", "publisher": "Microsoft Source", "url": "https://news.microsoft.com/source/2026/07/08/microsoft-announces-quarterly-earnings-release-date-68/", "published": "2026-07-08; results scheduled 2026-07-29", "retrieved": "2026-07-29T19:17:00-04:00", "type": "Primary" }, "mcp-releases-jul29": { "id": "mcp-releases-jul29", "title": "Releases · modelcontextprotocol/modelcontextprotocol", "publisher": "Model Context Protocol on GitHub", "url": "https://github.com/modelcontextprotocol/modelcontextprotocol/releases", "published": "Status at 2026-07-29 cutoff", "retrieved": "2026-07-29T19:20:00-04:00", "type": "Official repository" }, "mcp-versioning-jul29": { "id": "mcp-versioning-jul29", "title": "Versioning - Model Context Protocol", "publisher": "Model Context Protocol", "url": "https://modelcontextprotocol.io/docs/learn/versioning", "published": "Status at 2026-07-29 cutoff", "retrieved": "2026-07-29T19:21:00-04:00", "type": "Official documentation" }, "eu-ai-transparency-guidelines-jul20": { "id": "eu-ai-transparency-guidelines-jul20", "title": "Commission publishes guidelines on transparency obligations for providers and deployers of certain AI systems", "publisher": "European Commission", "url": "https://digital-strategy.ec.europa.eu/en/news/commission-publishes-guidelines-transparency-obligations-providers-and-deployers-certain-ai-systems", "published": "2026-07-20", "retrieved": "2026-07-29T19:24:00-04:00", "type": "Primary" }, "microsoft-project-perception-jul27": { "id": "microsoft-project-perception-jul27", "title": "Rethinking security for the age of AI", "publisher": "Microsoft", "url": "https://blogs.microsoft.com/blog/2026/07/27/rethinking-security-for-the-age-of-ai/", "published": "2026-07-27", "retrieved": "2026-07-29T19:27:00-04:00", "type": "Primary" }, "market-spx-jul29": { "id": "market-spx-jul29", "title": "S&P 500 Index historical data", "publisher": "MarketWatch", "url": "https://www.marketwatch.com/investing/index/spx/historical", "published": "2026-07-29 close", "retrieved": "2026-07-29T19:30:00-04:00", "type": "Historical market data" }, "market-dji-jul29": { "id": "market-dji-jul29", "title": "Dow Jones Industrial Average historical data", "publisher": "Yahoo Finance", "url": "https://finance.yahoo.com/quote/%5EDJI/history/", "published": "2026-07-29 close", "retrieved": "2026-07-29T19:31:00-04:00", "type": "Historical market data" }, "market-ixic-jul29": { "id": "market-ixic-jul29", "title": "Nasdaq Composite historical data", "publisher": "Yahoo Finance", "url": "https://finance.yahoo.com/quote/%5EIXIC/history/", "published": "2026-07-29 close", "retrieved": "2026-07-29T19:32:00-04:00", "type": "Historical market data" }, "reuters-amazon-aws-jul30": { "id": "reuters-amazon-aws-jul30", "title": "Amazon lifts investment plans after strong cloud sales; shares jump", "publisher": "Reuters", "url": "https://www.reuters.com/business/retail-consumer/amazon-beats-estimates-quarterly-cloud-revenue-growth-2026-07-30/", "published": "2026-07-30", "retrieved": "2026-07-30T21:50:00-04:00", "type": "Market and earnings reporting" }, "reuters-wall-street-microsoft-jul30": { "id": "reuters-wall-street-microsoft-jul30", "title": "Wall Street ends sharply higher, lifted by soaring Microsoft", "publisher": "Reuters", "url": "https://www.reuters.com/business/us-stock-futures-steady-microsoft-offsets-fed-concerns-2026-07-30/", "published": "2026-07-30 close", "retrieved": "2026-07-30T21:53:00-04:00", "type": "Market and earnings reporting" }, "ap-apple-q3-jul30": { "id": "ap-apple-q3-jul30", "title": "Strong iPhone sales propel Apple fiscal Q3 results past Wall Street expectations", "publisher": "The Associated Press", "url": "https://apnews.com/article/94102918cb3592ebc1d2a38c4d7d819a", "published": "2026-07-30", "retrieved": "2026-07-30T21:56:00-04:00", "type": "Company earnings reporting" }, "amazon-q2-event-jul30": { "id": "amazon-q2-event-jul30", "title": "Q2 2026 Amazon.com earnings conference call", "publisher": "Amazon Investor Relations", "url": "https://ir.aboutamazon.com/events/event-details/default.aspx", "published": "2026-07-30", "retrieved": "2026-07-30T21:58:00-04:00", "type": "Official company event" }, "mcp-releases-jul30": { "id": "mcp-releases-jul30", "title": "Releases · modelcontextprotocol/modelcontextprotocol", "publisher": "Model Context Protocol on GitHub", "url": "https://github.com/modelcontextprotocol/modelcontextprotocol/releases", "published": "Stable 2026-07-28 release verified 2026-07-30", "retrieved": "2026-07-30T22:01:00-04:00", "type": "Official repository" }, "mcp-versioning-jul30": { "id": "mcp-versioning-jul30", "title": "Versioning - Model Context Protocol 2026-07-28", "publisher": "Model Context Protocol", "url": "https://modelcontextprotocol.io/docs/2026-07-28/learn/versioning", "published": "Current documentation verified 2026-07-30", "retrieved": "2026-07-30T22:03:00-04:00", "type": "Official documentation" }, "market-spx-jul30": { "id": "market-spx-jul30", "title": "S&P 500 Index historical data", "publisher": "MarketWatch", "url": "https://www.marketwatch.com/investing/index/spx/historical", "published": "2026-07-30 close", "retrieved": "2026-07-30T22:05:00-04:00", "type": "Historical market data" }, "market-dji-jul30": { "id": "market-dji-jul30", "title": "Dow Jones Industrial Average historical data", "publisher": "Yahoo Finance", "url": "https://finance.yahoo.com/quote/%5EDJI/history/", "published": "2026-07-30 close", "retrieved": "2026-07-30T22:07:00-04:00", "type": "Historical market data" }, "market-ixic-jul30": { "id": "market-ixic-jul30", "title": "Nasdaq Composite historical data", "publisher": "Yahoo Finance", "url": "https://finance.yahoo.com/quote/%5EIXIC/history/", "published": "2026-07-30 close", "retrieved": "2026-07-30T22:09:00-04:00", "type": "Historical market data" }, "eu-article50-aug2": { "id": "eu-article50-aug2", "title": "Commission publishes guidelines on transparency obligations for providers and deployers of certain AI systems", "publisher": "European Commission", "url": "https://digital-strategy.ec.europa.eu/en/news/commission-publishes-guidelines-transparency-obligations-providers-and-deployers-certain-ai-systems", "published": "2026-07-20; obligations apply 2026-08-02", "retrieved": "2026-08-02T21:45:00-04:00", "type": "Official regulatory guidance" }, "eu-gpai-enforcement-aug2": { "id": "eu-gpai-enforcement-aug2", "title": "Commission enforcement powers for obligations covering the most advanced general-purpose AI models", "publisher": "European Commission AI Act Service Desk", "url": "https://ai-act-service-desk.ec.europa.eu/en/faq?faq_category_id=69", "published": "Enforcement powers apply 2026-08-02", "retrieved": "2026-08-02T21:49:00-04:00", "type": "Official enforcement guidance" }, "eu-ai-act-timeline-aug2": { "id": "eu-ai-act-timeline-aug2", "title": "Timeline for the implementation of the EU AI Act", "publisher": "European Commission AI Act Service Desk", "url": "https://ai-act-service-desk.ec.europa.eu/en/ai-act/timeline/timeline-implementation-eu-ai-act", "published": "Current timeline verified 2026-08-02", "retrieved": "2026-08-02T21:52:00-04:00", "type": "Official implementation timeline" }, "eu-governance-enforcement-aug2": { "id": "eu-governance-enforcement-aug2", "title": "Governance and enforcement of the AI Act", "publisher": "European Commission", "url": "https://digital-strategy.ec.europa.eu/en/policies/ai-act-governance-and-enforcement", "published": "Current governance page verified 2026-08-02", "retrieved": "2026-08-02T21:55:00-04:00", "type": "Official governance guidance" }, "huggingface-agent-incident-jul16": { "id": "huggingface-agent-incident-jul16", "title": "Security incident disclosure — July 2026", "publisher": "Hugging Face", "url": "https://huggingface.co/blog/security-incident-july-2026", "published": "2026-07-16", "retrieved": "2026-08-02T21:58:00-04:00", "type": "Official security disclosure" }, "reuters-eu-agent-talks-jul31": { "id": "reuters-eu-agent-talks-jul31", "title": "EU in talks with OpenAI, Anthropic after rogue AI agent hacks", "publisher": "Reuters", "url": "https://www.reuters.com/world/eu-says-necessary-monitor-high-risk-ai-systems-after-openai-anthropic-ai-hacking-2026-07-31/", "published": "2026-07-31", "retrieved": "2026-08-02T22:04:00-04:00", "type": "Regulatory and security reporting" }, "reuters-us-voluntary-tests-aug3": { "id": "reuters-us-voluntary-tests-aug3", "title": "Meta, Anthropic, Google, OpenAI to meet Trump officials about AI safety testing", "publisher": "Reuters", "url": "https://www.reuters.com/world/us-finalizes-voluntary-ai-safety-tests-white-house-official-says-2026-08-03/", "published": "2026-08-03", "retrieved": "2026-08-04T08:38:00-04:00", "type": "Government and technology reporting" }, "whitehouse-frontier-security-order-jun2": { "id": "whitehouse-frontier-security-order-jun2", "title": "Promoting Advanced Artificial Intelligence Innovation and Security", "publisher": "The White House", "url": "https://www.whitehouse.gov/presidential-actions/2026/06/promoting-advanced-artificial-intelligence-innovation-and-security/", "published": "2026-06-02", "retrieved": "2026-08-04T08:42:00-04:00", "type": "Official executive order" }, "reuters-uk-voluntary-regulation-aug3": { "id": "reuters-uk-voluntary-regulation-aug3", "title": "Britain says it is open to AI regulation if voluntary safeguards fall short", "publisher": "Reuters", "url": "https://www.reuters.com/business/media-telecom/britain-says-it-is-open-ai-regulation-if-voluntary-safeguards-fall-short-2026-08-03/", "published": "2026-08-03", "retrieved": "2026-08-04T08:46:00-04:00", "type": "Government and technology reporting" }, "uk-ai-cyber-code-jul10": { "id": "uk-ai-cyber-code-jul10", "title": "Code of Practice for the Cyber Security of AI", "publisher": "UK Government", "url": "https://www.gov.uk/government/publications/ai-cyber-security-code-of-practice/code-of-practice-for-the-cyber-security-of-ai", "published": "Current code verified 2026-08-04", "retrieved": "2026-08-04T08:49:00-04:00", "type": "Official cyber-security guidance" }, "reuters-palantir-q2-aug3": { "id": "reuters-palantir-q2-aug3", "title": "Palantir lifts annual revenue forecast on steady demand for AI-powered data analytics", "publisher": "Reuters", "url": "https://www.reuters.com/technology/palantir-raises-annual-revenue-forecast-strong-demand-us-government-commercial-2026-08-03/", "published": "2026-08-03", "retrieved": "2026-08-04T08:53:00-04:00", "type": "Company earnings reporting" }, "ap-us-markets-aug3": { "id": "ap-us-markets-aug3", "title": "How major US stock indexes fared Monday 8/3/2026", "publisher": "The Associated Press", "url": "https://apnews.com/article/wall-street-stocks-dow-nasdaq-f8e5f81b45c83878f5b7f69832bb0c95", "published": "2026-08-03 close", "retrieved": "2026-08-04T08:57:00-04:00", "type": "Market close reporting" }, "reuters-ai-futures-aug4": { "id": "reuters-ai-futures-aug4", "title": "US stock futures advance as strong AI-linked earnings continue", "publisher": "Reuters", "url": "https://www.reuters.com/business/nasdaq-futures-underpinned-by-strong-ai-forecasts-focus-earnings-data-2026-08-04/", "published": "2026-08-04", "retrieved": "2026-08-04T09:03:00-04:00", "type": "Premarket and earnings reporting" }, "market-spx-aug3": { "id": "market-spx-aug3", "title": "S&P 500 Index historical data", "publisher": "MarketWatch", "url": "https://www.marketwatch.com/investing/index/spx/historical", "published": "2026-08-03 close", "retrieved": "2026-08-04T10:54:00-04:00", "type": "Historical market data" }, "market-dji-aug3": { "id": "market-dji-aug3", "title": "Dow Jones Industrial Average historical data", "publisher": "Yahoo Finance", "url": "https://finance.yahoo.com/quote/%5EDJI/history/", "published": "2026-08-03 close", "retrieved": "2026-08-04T10:55:00-04:00", "type": "Historical market data" }, "market-ixic-aug3": { "id": "market-ixic-aug3", "title": "Nasdaq Composite historical data", "publisher": "Yahoo Finance", "url": "https://finance.yahoo.com/quote/%5EIXIC/history/", "published": "2026-08-03 close", "retrieved": "2026-08-04T10:56:00-04:00", "type": "Historical market data" }, "reuters-whitehouse-open-weight-aug4": { "id": "reuters-whitehouse-open-weight-aug4", "title": "Trump advisers tell AI firms they will not safety-test open-weight models", "publisher": "Reuters", "url": "https://www.reuters.com/legal/litigation/meta-anthropic-google-openai-meet-with-trump-white-house-amid-rogue-ai-agent-2026-08-04/", "published": "2026-08-04", "retrieved": "2026-08-04T20:52:00-04:00", "type": "Government and technology reporting" }, "aisi-unsanctioned-agent-report-aug4": { "id": "aisi-unsanctioned-agent-report-aug4", "title": "Incident Report: unsanctioned agent behaviour during cyber testing", "publisher": "UK AI Security Institute", "url": "https://www.aisi.gov.uk/blog/incident-report-unsanctioned-agent-behaviour-during-cyber-testing", "published": "2026-08-04", "retrieved": "2026-08-04T20:56:00-04:00", "type": "Official government incident report" }, "openai-third-party-cyber-evals-aug4": { "id": "openai-third-party-cyber-evals-aug4", "title": "Third-party cyber evaluations involving OpenAI models", "publisher": "OpenAI", "url": "https://openai.com/index/third-party-cyber-evaluations-involving-openai-models/", "published": "2026-08-04", "retrieved": "2026-08-04T21:01:00-04:00", "type": "Official company security disclosure" }, "reuters-agent-breaches-aug5": { "id": "reuters-agent-breaches-aug5", "title": "OpenAI, Anthropic AI agents implicated in new security breaches", "publisher": "Reuters", "url": "https://www.reuters.com/legal/litigation/openai-anthropic-ai-agents-implicated-new-security-breaches-2026-08-05/", "published": "2026-08-05", "retrieved": "2026-08-04T21:06:00-04:00", "type": "Cybersecurity and technology reporting" }, "reuters-anthropic-global-affairs-aug4": { "id": "reuters-anthropic-global-affairs-aug4", "title": "Anthropic names global affairs chief to tackle AI policy as Trump tensions persist", "publisher": "Reuters", "url": "https://www.reuters.com/world/anthropic-names-global-affairs-chief-tackle-ai-policy-trump-tensions-persist-2026-08-04/", "published": "2026-08-04", "retrieved": "2026-08-04T21:11:00-04:00", "type": "Company and policy reporting" }, "reuters-us-markets-aug4": { "id": "reuters-us-markets-aug4", "title": "Dow, S&P 500 close at record on AI-linked earnings, Mideast deal hopes", "publisher": "Reuters", "url": "https://www.reuters.com/business/nasdaq-futures-underpinned-by-strong-ai-forecasts-focus-earnings-data-2026-08-04/", "published": "2026-08-04 close", "retrieved": "2026-08-04T21:16:00-04:00", "type": "Market close and company earnings reporting" }, "market-spx-aug4": { "id": "market-spx-aug4", "title": "S&P 500 Index historical data — August 4, 2026 close", "publisher": "MarketWatch", "url": "https://www.marketwatch.com/investing/index/spx/historical", "published": "2026-08-04 close", "retrieved": "2026-08-04T21:20:00-04:00", "type": "Historical market data" }, "market-dji-aug4": { "id": "market-dji-aug4", "title": "Dow Jones Industrial Average historical data — August 4, 2026 close", "publisher": "Yahoo Finance", "url": "https://finance.yahoo.com/quote/%5EDJI/history/", "published": "2026-08-04 close", "retrieved": "2026-08-04T21:21:00-04:00", "type": "Historical market data" }, "market-ixic-aug4": { "id": "market-ixic-aug4", "title": "Nasdaq Composite historical data — August 4, 2026 close", "publisher": "Yahoo Finance", "url": "https://finance.yahoo.com/quote/%5EIXIC/history/", "published": "2026-08-04 close", "retrieved": "2026-08-04T21:22:00-04:00", "type": "Historical market data" }, "reuters-google-ai-leadership-aug5": { "id": "reuters-google-ai-leadership-aug5", "title": "Google shakes up AI leadership as DeepMind chief shifts role", "publisher": "Reuters", "url": "https://www.reuters.com/business/google-shakes-up-ai-leadership-deepmind-chief-shifts-role-2026-08-05/", "published": "2026-08-05", "retrieved": "2026-08-05T20:42:00-04:00", "type": "Company and technology reporting" }, "reuters-meta-cyber-eval-aug5": { "id": "reuters-meta-cyber-eval-aug5", "title": "Meta AI model hacks another company during testing", "publisher": "Reuters", "url": "https://www.reuters.com/technology/metas-ai-model-hacked-another-company-during-testing-information-reports-2026-08-05/", "published": "2026-08-05", "retrieved": "2026-08-05T20:47:00-04:00", "type": "Cybersecurity and technology reporting" }, "reuters-amd-ai-payoff-aug5": { "id": "reuters-amd-ai-payoff-aug5", "title": "AMD falls as investors demand bigger AI payoff", "publisher": "Reuters", "url": "https://www.reuters.com/business/amd-falls-investors-seek-bigger-ai-payoff-2026-08-05/", "published": "2026-08-05", "retrieved": "2026-08-05T20:52:00-04:00", "type": "Semiconductor and market reporting" }, "reuters-global-markets-aug5": { "id": "reuters-global-markets-aug5", "title": "Stocks mixed with Nasdaq down; US oil eases on Iran talks", "publisher": "Reuters", "url": "https://www.reuters.com/world/china/global-markets-global-markets-2026-08-05/", "published": "2026-08-05 close", "retrieved": "2026-08-05T20:57:00-04:00", "type": "Market close reporting" }, "market-spx-aug5": { "id": "market-spx-aug5", "title": "S&P 500 Index historical data — August 5, 2026 close", "publisher": "MarketWatch", "url": "https://www.marketwatch.com/investing/index/spx/historical", "published": "2026-08-05 close", "retrieved": "2026-08-05T21:02:00-04:00", "type": "Historical market data" }, "market-dji-aug5": { "id": "market-dji-aug5", "title": "Dow Jones Industrial Average historical data — August 5, 2026 close", "publisher": "Yahoo Finance", "url": "https://finance.yahoo.com/quote/%5EDJI/history/", "published": "2026-08-05 close", "retrieved": "2026-08-05T21:03:00-04:00", "type": "Historical market data" }, "market-ixic-aug5": { "id": "market-ixic-aug5", "title": "Nasdaq Composite historical data — August 5, 2026 close", "publisher": "Yahoo Finance", "url": "https://finance.yahoo.com/quote/%5EIXIC/history/", "published": "2026-08-05 close", "retrieved": "2026-08-05T21:04:00-04:00", "type": "Historical market data" }, "reuters-alibaba-qwen38-license-aug7": { "id": "reuters-alibaba-qwen38-license-aug7", "title": "Alibaba plans to charge big users of its next open-source AI model, sources say", "publisher": "Reuters", "url": "https://www.reuters.com/business/retail-consumer/alibaba-plans-charge-big-users-its-next-open-source-ai-model-sources-say-2026-08-07/", "published": "2026-08-07", "retrieved": "2026-08-07T02:18:00-04:00", "type": "Company, licensing and AI-market reporting" }, "reuters-amd-taalas-aug6": { "id": "reuters-amd-taalas-aug6", "title": "AMD deepens AI inference bet with Taalas deal as chip race heats up", "publisher": "Reuters", "url": "https://www.reuters.com/business/amd-deepens-ai-inference-bet-with-taalas-deal-chip-race-heats-up-2026-08-06/", "published": "2026-08-06", "retrieved": "2026-08-07T02:21:00-04:00", "type": "Semiconductor and acquisition reporting" }, "reuters-deepseek-unitree-aug6": { "id": "reuters-deepseek-unitree-aug6", "title": "DeepSeek invests $20.8 million in Unitree's Shanghai IPO", "publisher": "Reuters", "url": "https://www.reuters.com/world/asia-pacific/deepseek-invests-208-million-unitrees-shanghai-ipo-2026-08-06/", "published": "2026-08-06", "retrieved": "2026-08-07T02:24:00-04:00", "type": "AI, robotics and China technology reporting" }, "reuters-deepseek-funding-aug6": { "id": "reuters-deepseek-funding-aug6", "title": "DeepSeek resumes funding round seeking nearly $8 billion, Bloomberg News reports", "publisher": "Reuters", "url": "https://www.reuters.com/world/asia-pacific/deepseek-resumes-funding-round-seeking-nearly-8-billion-bloomberg-news-reports-2026-08-06/", "published": "2026-08-06", "retrieved": "2026-08-07T02:27:00-04:00", "type": "Company funding and valuation reporting" }, "reuters-us-ai-policy-pressure-aug6": { "id": "reuters-us-ai-policy-pressure-aug6", "title": "Trump's tech ties come under bipartisan fire after AI agents go rogue", "publisher": "Reuters", "url": "https://www.reuters.com/legal/litigation/trumps-tech-ties-blamed-ai-inaction-by-maga-republicans-democrats-2026-08-06/", "published": "2026-08-06", "retrieved": "2026-08-07T02:30:00-04:00", "type": "AI policy and cybersecurity reporting" }, "reuters-us-markets-aug6": { "id": "reuters-us-markets-aug6", "title": "Stocks close lower as investors eye Mideast talks, earnings", "publisher": "Reuters", "url": "https://www.reuters.com/business/retail-consumer/sp-500-dow-futures-steady-mideast-deal-focus-chips-stumble-2026-08-06/", "published": "2026-08-06 close", "retrieved": "2026-08-07T02:33:00-04:00", "type": "Market close and company earnings reporting" }, "reuters-global-markets-aug7": { "id": "reuters-global-markets-aug7", "title": "Asian shares pause for US jobs, oil extends gains on Mideast risk", "publisher": "Reuters", "url": "https://www.reuters.com/world/china/global-markets-global-markets-2026-08-07/", "published": "2026-08-07", "retrieved": "2026-08-07T02:36:00-04:00", "type": "Global market and macroeconomic reporting" }, "market-spx-aug6": { "id": "market-spx-aug6", "title": "S&P 500 Index historical data — August 6, 2026 close", "publisher": "MarketWatch", "url": "https://www.marketwatch.com/investing/index/spx/historical", "published": "2026-08-06 close", "retrieved": "2026-08-07T02:39:00-04:00", "type": "Historical market data" }, "market-dji-aug6": { "id": "market-dji-aug6", "title": "Dow Jones Industrial Average historical data — August 6, 2026 close", "publisher": "Yahoo Finance", "url": "https://finance.yahoo.com/quote/%5EDJI/history/", "published": "2026-08-06 close", "retrieved": "2026-08-07T02:40:00-04:00", "type": "Historical market data" }, "market-ixic-aug6": { "id": "market-ixic-aug6", "title": "Nasdaq Composite historical data — August 6, 2026 close", "publisher": "Yahoo Finance", "url": "https://finance.yahoo.com/quote/%5EIXIC/history/", "published": "2026-08-06 close", "retrieved": "2026-08-07T02:41:00-04:00", "type": "Historical market data" }, "whitehouse-frontier-ai-eo-jun2": { "id": "whitehouse-frontier-ai-eo-jun2", "title": "Promoting Advanced Artificial Intelligence Innovation and Security", "publisher": "The White House", "url": "https://www.whitehouse.gov/presidential-actions/2026/06/promoting-advanced-artificial-intelligence-innovation-and-security/", "published": "2026-06-02", "retrieved": "2026-08-08T00:52:00-04:00", "type": "Primary government policy source" }, "guardian-whitehouse-ai-framework-aug7": { "id": "guardian-whitehouse-ai-framework-aug7", "title": "The White House's plan to vet potentially dangerous AI is cloaked in secrecy", "publisher": "The Guardian", "url": "https://www.theguardian.com/technology/2026/aug/07/white-house-ai", "published": "2026-08-07", "retrieved": "2026-08-08T00:57:00-04:00", "type": "AI policy and transparency reporting" }, "reuters-open-weight-testing-aug4": { "id": "reuters-open-weight-testing-aug4", "title": "Trump advisers tell AI firms they will not safety-test open-weight models", "publisher": "Reuters", "url": "https://www.reuters.com/legal/litigation/meta-anthropic-google-openai-meet-with-trump-white-house-amid-rogue-ai-agent-2026-08-04/", "published": "2026-08-04", "retrieved": "2026-08-08T01:02:00-04:00", "type": "AI policy and cybersecurity reporting" }, "reuters-us-jobs-aug7": { "id": "reuters-us-jobs-aug7", "title": "US suffers unexpected job losses in July, markets dial back rate hike expectations", "publisher": "Reuters", "url": "https://www.reuters.com/business/us-nonfarm-payrolls-fall-july-unemployment-rate-eases-41-2026-08-07/", "published": "2026-08-07", "retrieved": "2026-08-08T01:07:00-04:00", "type": "U.S. labor-market and macroeconomic reporting" }, "reuters-us-markets-aug7": { "id": "reuters-us-markets-aug7", "title": "S&P closes at record high as soft jobs report eases rate-hike concerns", "publisher": "Reuters", "url": "https://www.reuters.com/business/sp-500-dow-futures-muted-ahead-jobs-data-chips-software-stocks-rise-2026-08-07/", "published": "2026-08-07 close", "retrieved": "2026-08-08T01:12:00-04:00", "type": "U.S. market close and rate-expectation reporting" }, "reuters-inflation-weekahead-aug7": { "id": "reuters-inflation-weekahead-aug7", "title": "Wall St Week Ahead: Inflation data to test record-setting US stocks, Fed rate views", "publisher": "Reuters", "url": "https://www.reuters.com/business/wall-st-week-ahead-inflation-data-test-record-setting-us-stocks-fed-rate-views-2026-08-07/", "published": "2026-08-07", "retrieved": "2026-08-08T01:17:00-04:00", "type": "Markets, inflation and technology-sector reporting" }, "wsj-source-foundry-aug7": { "id": "wsj-source-foundry-aug7", "title": "Situational Awareness Bets $400 Million on Stealth Chip Startup After Crash", "publisher": "The Wall Street Journal", "url": "https://www.wsj.com/tech/ai/situational-awareness-bets-400-million-on-stealth-chip-startup-after-crash-02c7374e", "published": "2026-08-07", "retrieved": "2026-08-08T01:22:00-04:00", "type": "AI infrastructure and semiconductor investment reporting" }, "market-dji-aug7": { "id": "market-dji-aug7", "title": "Dow Jones Industrial Average historical data — August 7, 2026 close", "publisher": "Yahoo Finance", "url": "https://finance.yahoo.com/quote/%5EDJI/history/", "published": "2026-08-07 close", "retrieved": "2026-08-08T01:27:00-04:00", "type": "Historical market data" }, "market-ixic-aug7": { "id": "market-ixic-aug7", "title": "Nasdaq Composite historical data — August 7, 2026 close", "publisher": "Yahoo Finance", "url": "https://finance.yahoo.com/quote/%5EIXIC/history/", "published": "2026-08-07 close", "retrieved": "2026-08-08T01:28:00-04:00", "type": "Historical market data" }, "market-spx-aug7": { "id": "market-spx-aug7", "title": "S&P 500 Index historical data — August 7, 2026 close", "publisher": "MarketWatch", "url": "https://www.marketwatch.com/investing/index/spx/historical", "published": "2026-08-07 close", "retrieved": "2026-08-08T01:26:00-04:00", "type": "Historical market data" }, "reuters-apple-qwen-mac-aug8": { "id": "reuters-apple-qwen-mac-aug8", "title": "Apple says Mac users in China can connect to Alibaba's Qwen AI service", "publisher": "Reuters", "url": "https://www.reuters.com/business/retail-consumer/apple-says-mac-users-china-can-connect-alibabas-qwen-ai-service-2026-08-08/", "published": "2026-08-08", "retrieved": "2026-08-08T19:10:00-04:00", "type": "AI product and China-market reporting" }, "reuters-openai-astra-cyber-aug7": { "id": "reuters-openai-astra-cyber-aug7", "title": "OpenAI flags possible critical cybersecurity risk in upcoming model, tightens controls", "publisher": "Reuters", "url": "https://www.reuters.com/legal/litigation/openai-flags-possible-critical-cybersecurity-risk-upcoming-model-tightens-2026-08-07/", "published": "2026-08-07", "retrieved": "2026-08-08T19:12:00-04:00", "type": "Frontier-model cybersecurity reporting" }, "openai-preparedness-framework-v2": { "id": "openai-preparedness-framework-v2", "title": "Preparedness Framework v2", "publisher": "OpenAI", "url": "https://cdn.openai.com/pdf/18a02b5d-6b67-4cec-ab64-68cdfbddebcd/preparedness-framework-v2.pdf", "published": "2026", "retrieved": "2026-08-08T19:14:00-04:00", "type": "Primary AI safety framework" }, "reuters-aisi-agent-evals-aug5": { "id": "reuters-aisi-agent-evals-aug5", "title": "OpenAI, Anthropic AI agents implicated in new security breaches", "publisher": "Reuters", "url": "https://www.reuters.com/legal/litigation/openai-anthropic-ai-agents-implicated-new-security-breaches-2026-08-05/", "published": "2026-08-05", "retrieved": "2026-08-08T19:16:00-04:00", "type": "AI agent evaluation and cybersecurity reporting" }, "reuters-nist-ai-evals-aug7": { "id": "reuters-nist-ai-evals-aug7", "title": "Trump says Congress wants to regulate AI industry 'out of business'", "publisher": "Reuters", "url": "https://www.reuters.com/legal/litigation/trump-says-congress-wants-regulate-ai-industry-out-business-2026-08-07/", "published": "2026-08-07", "retrieved": "2026-08-08T19:18:00-04:00", "type": "U.S. AI policy and standards reporting" }, "apple-cn-intelligence-support-aug8": { "id": "apple-cn-intelligence-support-aug8", "title": "Use Apple Intelligence on Mac", "publisher": "Apple Support China", "url": "https://support.apple.com/zh-cn/guide/mac-help-cn/mchl46361784/26/mac/26", "published": "Current support documentation", "retrieved": "2026-08-08T19:20:00-04:00", "type": "Primary product support documentation" }, "openai-atlas-retirement-aug9": { "id": "openai-atlas-retirement-aug9", "title": "Evolving Atlas into ChatGPT for browser-based agentic work", "publisher": "OpenAI Help Center", "url": "https://help.openai.com/en/articles/20001371-evolving-atlas-into-chatgpt-for-browser-based-agentic-work", "published": "2026-08-09 retirement date", "retrieved": "2026-08-09T21:18:00-04:00", "type": "Primary product documentation" }, "anthropic-opus5-release-jul24": { "id": "anthropic-opus5-release-jul24", "title": "Claude Platform release notes — Claude Opus 5", "publisher": "Anthropic", "url": "https://platform.claude.com/docs/en/release-notes/overview", "published": "2026-07-24", "retrieved": "2026-08-09T21:22:00-04:00", "type": "Primary model documentation" }, "arcprize-opus5-jul24": { "id": "arcprize-opus5-jul24", "title": "Claude Opus 5 — ARC-AGI results", "publisher": "ARC Prize", "url": "https://arcprize.org/results/anthropic-claude-opus-5", "published": "2026-07-24", "retrieved": "2026-08-09T21:25:00-04:00", "type": "Benchmark-owner result page" }, "arcprize-gpt56-jul9": { "id": "arcprize-gpt56-jul9", "title": "GPT-5.6 — ARC-AGI results", "publisher": "ARC Prize", "url": "https://arcprize.org/results/openai-gpt-5-6", "published": "2026-07-09", "retrieved": "2026-08-09T21:27:00-04:00", "type": "Benchmark-owner result page" }, "reuters-qwen38-aug3": { "id": "reuters-qwen38-aug3", "title": "Alibaba unveils Qwen3.8-Max", "publisher": "Reuters", "url": "https://www.reuters.com/business/retail-consumer/alibaba-unveils-its-most-capable-ai-model-date-not-far-behind-moonshots-size-2026-08-03/", "published": "2026-08-03", "retrieved": "2026-08-09T21:31:00-04:00", "type": "Model release reporting" }, "alibaba-model-catalog-aug9": { "id": "alibaba-model-catalog-aug9", "title": "Alibaba Cloud Model Studio model catalog", "publisher": "Alibaba Cloud", "url": "https://www.alibabacloud.com/help/en/model-studio/models", "published": "Current documentation", "retrieved": "2026-08-09T21:33:00-04:00", "type": "Primary model catalog" }, "alibaba-model-pricing-aug9": { "id": "alibaba-model-pricing-aug9", "title": "Alibaba Cloud Model Studio pricing", "publisher": "Alibaba Cloud", "url": "https://www.alibabacloud.com/help/en/model-studio/model-pricing", "published": "Current documentation", "retrieved": "2026-08-09T21:35:00-04:00", "type": "Primary pricing documentation" }, "deepseek-v4-apr24": { "id": "deepseek-v4-apr24", "title": "DeepSeek V4-Pro and V4-Flash release notes", "publisher": "DeepSeek", "url": "https://api-docs.deepseek.com/news/news260424/", "published": "2026-04-24", "retrieved": "2026-08-09T21:38:00-04:00", "type": "Primary model release documentation" }, "deepseek-v4-pricing-aug9": { "id": "deepseek-v4-pricing-aug9", "title": "DeepSeek API pricing — V4-Flash", "publisher": "DeepSeek", "url": "https://api-docs.deepseek.com/quick_start/pricing/", "published": "Current documentation", "retrieved": "2026-08-09T21:40:00-04:00", "type": "Primary API pricing documentation" }, "artificialanalysis-v4flash-aug9": { "id": "artificialanalysis-v4flash-aug9", "title": "DeepSeek V4 Flash — independent model analysis", "publisher": "Artificial Analysis", "url": "https://artificialanalysis.ai/models/deepseek-v4-flash", "published": "Current evaluation page", "retrieved": "2026-08-09T21:43:00-04:00", "type": "Independent model evaluation" }, "reuters-deepseek-v4-aug3": { "id": "reuters-deepseek-v4-aug3", "title": "DeepSeek V4-Flash cost and performance reporting", "publisher": "Reuters", "url": "https://www.reuters.com/business/retail-consumer/deepseeks-new-ai-model-is-by-far-cheapest-well-known-models-run-research-firm-2026-08-03/", "published": "2026-08-03", "retrieved": "2026-08-09T21:46:00-04:00", "type": "Model economics reporting" }, "reuters-bigtech-leases-aug4": { "id": "reuters-bigtech-leases-aug4", "title": "AI data-centre race builds $1 trillion lease burden for Big Tech", "publisher": "Reuters", "url": "https://www.reuters.com/business/retail-consumer/ai-data-centre-race-builds-1-trillion-lease-burden-big-tech-2026-08-04/", "published": "2026-08-04", "retrieved": "2026-08-09T21:50:00-04:00", "type": "Infrastructure and financial reporting" }, "sec-msft-leases-mar31": { "id": "sec-msft-leases-mar31", "title": "Microsoft Form 10-Q — leases not yet commenced", "publisher": "U.S. Securities and Exchange Commission", "url": "https://www.sec.gov/Archives/edgar/data/789019/000119312526191507/msft-20260331.htm", "published": "2026-03-31 quarter", "retrieved": "2026-08-09T21:53:00-04:00", "type": "Primary financial filing" }, "sec-meta-leases-mar31": { "id": "sec-meta-leases-mar31", "title": "Meta Platforms Form 10-Q lease commitments", "publisher": "U.S. Securities and Exchange Commission", "url": "https://www.sec.gov/Archives/edgar/data/1326801/000162828026028526/R16.htm", "published": "2026-03-31 quarter", "retrieved": "2026-08-09T21:56:00-04:00", "type": "Primary financial filing" }, "sec-alphabet-leases-mar31": { "id": "sec-alphabet-leases-mar31", "title": "Alphabet Form 10-Q — leases not yet commenced", "publisher": "U.S. Securities and Exchange Commission", "url": "https://www.sec.gov/Archives/edgar/data/1652044/000165204426000048/goog-20260331.htm", "published": "2026-03-31 quarter", "retrieved": "2026-08-09T21:58:00-04:00", "type": "Primary financial filing" }, "aisi-test-time-compute": { "id": "aisi-test-time-compute", "title": "Measuring AI agents progress on multi-step cyber attack scenarios", "publisher": "UK AI Security Institute", "url": "https://www.aisi.gov.uk/research/measuring-ai-agents-progress-on-multi-step-cyber-attack-scenarios", "published": "2026 research", "retrieved": "2026-08-09T22:02:00-04:00", "type": "Government AI evaluation research" }, "anthropic-models-overview-aug9": { "id": "anthropic-models-overview-aug9", "title": "Claude models overview", "publisher": "Anthropic", "url": "https://platform.claude.com/docs/en/about-claude/models/overview", "published": "Current documentation", "retrieved": "2026-08-09T22:28:00-04:00", "type": "Primary model documentation" }, "reuters-nvidia-financing-aug10": { "id": "reuters-nvidia-financing-aug10", "title": "Wall Street giants partner with Nvidia on $500 billion AI financing plan", "publisher": "Reuters", "url": "https://www.reuters.com/technology/wall-street-giants-partner-with-nvidia-500-billion-ai-financing-deal-ft-reports-2026-08-10/", "published": "2026-08-10", "retrieved": "2026-08-10T23:08:00-04:00", "type": "AI infrastructure financing reporting" }, "nvidia-naver-brookfield-jul24": { "id": "nvidia-naver-brookfield-jul24", "title": "NAVER, NVIDIA and Brookfield to Expand Korea’s National AI Factory Infrastructure Buildout", "publisher": "NVIDIA Newsroom", "url": "https://nvidianews.nvidia.com/news/naver-nvidia-and-brookfield-to-expand-koreas-national-ai-factory-infrastructure-buildout", "published": "2026-07-24", "retrieved": "2026-08-10T23:08:00-04:00", "type": "Primary infrastructure announcement" }, "reuters-datacenter-lender-risk-aug10": { "id": "reuters-datacenter-lender-risk-aug10", "title": "Lenders scrutinize U.S. data-center financing as community opposition builds", "publisher": "Reuters", "url": "https://www.reuters.com/legal/transactional/lenders-scrutinize-us-data-center-financing-community-opposition-builds-2026-08-10/", "published": "2026-08-10", "retrieved": "2026-08-10T23:08:00-04:00", "type": "Infrastructure financing and permitting reporting" }, "reuters-house-agent-oversight-aug10": { "id": "reuters-house-agent-oversight-aug10", "title": "U.S. House Democrats press Anthropic, OpenAI about AI-agent incidents", "publisher": "Reuters", "url": "https://www.reuters.com/legal/litigation/us-house-democrats-press-anthropic-openai-about-rogue-ai-agents-2026-08-10/", "published": "2026-08-10", "retrieved": "2026-08-10T23:08:00-04:00", "type": "U.S. policy and AI-agent safety reporting" }, "openai-hf-incident-jul21": { "id": "openai-hf-incident-jul21", "title": "Hugging Face model-evaluation security incident", "publisher": "OpenAI", "url": "https://openai.com/index/hugging-face-model-evaluation-security-incident/", "published": "2026-07-21", "retrieved": "2026-08-10T23:08:00-04:00", "type": "Primary security-incident disclosure" }, "huggingface-agent-intrusion-jul27": { "id": "huggingface-agent-intrusion-jul27", "title": "Agent intrusion technical timeline", "publisher": "Hugging Face", "url": "https://huggingface.co/blog/agent-intrusion-technical-timeline", "published": "2026-07-27", "retrieved": "2026-08-10T23:08:00-04:00", "type": "Primary security-incident reconstruction" }, "anthropic-containment-may25": { "id": "anthropic-containment-may25", "title": "How we contain Claude across products", "publisher": "Anthropic", "url": "https://www.anthropic.com/engineering/how-we-contain-claude", "published": "2026-05-25", "retrieved": "2026-08-10T23:08:00-04:00", "type": "Primary agent-containment engineering" }, "reuters-meta-glimmer-aug10": { "id": "reuters-meta-glimmer-aug10", "title": "Meta launches new AI model as Zuckerberg champions open-weight push", "publisher": "Reuters", "url": "https://www.reuters.com/world/china/meta-launches-new-ai-model-zuckerberg-champions-open-weight-push-2026-08-10/", "published": "2026-08-10", "retrieved": "2026-08-10T23:08:00-04:00", "type": "Model release and strategy reporting" }, "reuters-maia300-aug10": { "id": "reuters-maia300-aug10", "title": "Microsoft plans to unveil its new Maia 300 AI chip this fall, The Information reports", "publisher": "Reuters", "url": "https://www.reuters.com/business/microsoft-plans-unveil-its-new-maia-300-ai-chip-this-fall-information-reports-2026-08-10/", "published": "2026-08-10", "retrieved": "2026-08-10T23:08:00-04:00", "type": "AI accelerator reporting" }, "microsoft-maia300-transcript-mar4": { "id": "microsoft-maia300-transcript-mar4", "title": "Morgan Stanley Technology, Media & Telecom Conference", "publisher": "Microsoft Investor Relations", "url": "https://www.microsoft.com/en-us/investor/events/fy-2026/morgan-stanley-tmt-conference", "published": "2026-03-04", "retrieved": "2026-08-10T23:08:00-04:00", "type": "Primary executive transcript" }, "cvpr-agentsafe-2026": { "id": "cvpr-agentsafe-2026", "title": "AGENTSAFE: Benchmarking the Safety of Embodied Agents on Hazardous Instructions", "publisher": "CVPR 2026", "url": "https://openaccess.thecvf.com/content/CVPR2026/html/Ying_AGENTSAFE_Benchmarking_the_Safety_of_Embodied_Agents_on_Hazardous_Instructions_CVPR_2026_paper.html", "published": "2026", "retrieved": "2026-08-10T23:08:00-04:00", "type": "Peer-reviewed conference paper" }, "acl-safeagent-2026": { "id": "acl-safeagent-2026", "title": "SafeAgent: Safeguarding LLM Agents via an Automated Risk Simulator", "publisher": "ACL 2026", "url": "https://aclanthology.org/2026.acl-long.1501/", "published": "2026", "retrieved": "2026-08-10T23:08:00-04:00", "type": "Peer-reviewed conference paper" }, "ap-us-markets-aug10": { "id": "ap-us-markets-aug10", "title": "How major U.S. stock indexes fared Monday 8/10/2026", "publisher": "Associated Press", "url": "https://apnews.com/article/37588f23787295860e468d19399b56ca", "published": "2026-08-10 close", "retrieved": "2026-08-10T23:08:00-04:00", "type": "Market close reporting" }, "reuters-coreweave-supermicro-aug12": { "id": "reuters-coreweave-supermicro-aug12", "title": "CoreWeave, Super Micro surge on signs of sustained AI buildout", "publisher": "Reuters", "url": "https://www.reuters.com/business/coreweave-super-micro-climb-signs-sustained-ai-buildout-2026-08-12/", "published": "2026-08-12", "retrieved": "2026-08-12T23:16:00-04:00", "type": "Current financial and infrastructure reporting" }, "coreweave-q1-backlog-definition-2026": { "id": "coreweave-q1-backlog-definition-2026", "title": "CoreWeave Reports Strong First Quarter 2026 Results", "publisher": "CoreWeave Investor Relations", "url": "https://investors.coreweave.com/news/news-details/2026/CoreWeave-Reports-Strong-First-Quarter-2026-Results/", "published": "2026-05-07", "retrieved": "2026-08-12T23:16:00-04:00", "type": "Primary company financial disclosure" }, "supermicro-q4-preliminary-jul21": { "id": "supermicro-q4-preliminary-jul21", "title": "Supermicro Provides Fourth Quarter of Fiscal Year 2026 Preliminary Business Update", "publisher": "Super Micro Computer Investor Relations", "url": "https://ir.supermicro.com/news/news-details/2026/Supermicro-Provides-Fourth-Quarter-of-Fiscal-Year-2026-Preliminary-Business-Update/default.aspx", "published": "2026-07-21", "retrieved": "2026-08-12T23:16:00-04:00", "type": "Primary company financial disclosure" }, "reuters-cisco-ai-orders-aug12": { "id": "reuters-cisco-ai-orders-aug12", "title": "Cisco forecasts annual revenue above estimates on sustained AI spending", "publisher": "Reuters", "url": "https://www.reuters.com/technology/cisco-forecasts-upbeat-annual-revenue-2026-08-12/", "published": "2026-08-12", "retrieved": "2026-08-12T23:16:00-04:00", "type": "Current financial and infrastructure reporting" }, "reuters-foxconn-cowos-aug12": { "id": "reuters-foxconn-cowos-aug12", "title": "Taiwan’s Foxconn reports 35% rise in Q2 profit on AI demand, beats forecasts", "publisher": "Reuters", "url": "https://www.reuters.com/world/asia-pacific/taiwans-foxconn-reports-35-rise-q2-profit-beats-forecasts-2026-08-12/", "published": "2026-08-12", "retrieved": "2026-08-12T23:16:00-04:00", "type": "Current manufacturing and AI-server reporting" }, "tsmc-cowos-3dfabric": { "id": "tsmc-cowos-3dfabric", "title": "3DFabric: CoWoS and advanced packaging", "publisher": "TSMC", "url": "https://3dfabric.tsmc.com/english/dedicatedFoundry/technology/3DFabric.htm", "published": "Current reference", "retrieved": "2026-08-12T23:16:00-04:00", "type": "Primary semiconductor technical documentation" }, "reuters-taiwan-ai-attack-aug13": { "id": "reuters-taiwan-ai-attack-aug13", "title": "Taiwan says it was targeted last month in AI-driven hacking campaign", "publisher": "Reuters", "url": "https://www.reuters.com/world/china/taiwan-says-it-was-targeted-last-month-ai-driven-hacking-campaign-2026-08-13/", "published": "2026-08-13 UTC / 2026-08-12 New York edition window", "retrieved": "2026-08-12T23:16:00-04:00", "type": "Government-confirmation reporting" }, "ft-taiwan-agent-cyber-aug12": { "id": "ft-taiwan-agent-cyber-aug12", "title": "China-linked hackers hit Taiwan in unprecedented autonomous AI cyber attack", "publisher": "Financial Times", "url": "https://www.ft.com/content/7d2ab3e0-9085-48f6-b38a-d90260d58795", "published": "2026-08-12", "retrieved": "2026-08-12T23:16:00-04:00", "type": "Investigative cyber reporting based on Dream findings" }, "taiwan-openclaw-advisory-mar2026": { "id": "taiwan-openclaw-advisory-mar2026", "title": "OpenClaw cybersecurity risk advisory", "publisher": "Taiwan Administration for Cyber Security", "url": "https://moda.gov.tw/ACS/press/news/press/19294", "published": "2026-03", "retrieved": "2026-08-12T23:16:00-04:00", "type": "Primary government security advisory" }, "reuters-google-deepmind-aug12": { "id": "reuters-google-deepmind-aug12", "title": "Inside the Google executive moves that led to its big AI reshuffle", "publisher": "Reuters", "url": "https://www.reuters.com/world/inside-google-executive-moves-that-led-its-big-ai-reshuffle-2026-08-12/", "published": "2026-08-12", "retrieved": "2026-08-12T23:16:00-04:00", "type": "Current organization and model-development reporting" }, "google-gemini-35-current": { "id": "google-gemini-35-current", "title": "Gemini 3.5", "publisher": "Google DeepMind", "url": "https://deepmind.google/models/gemini/", "published": "Current product page", "retrieved": "2026-08-12T23:16:00-04:00", "type": "Primary model availability source" }, "reuters-cerebras-q2-aug12": { "id": "reuters-cerebras-q2-aug12", "title": "Cerebras raises annual targets on strong AI chip demand", "publisher": "Reuters", "url": "https://www.reuters.com/business/cerebras-raises-annual-targets-strong-ai-chip-demand-2026-08-12/", "published": "2026-08-12", "retrieved": "2026-08-12T23:16:00-04:00", "type": "Current financial reporting" }, "reuters-lightcap-openai-aug11": { "id": "reuters-lightcap-openai-aug11", "title": "Senior OpenAI executive Brad Lightcap to leave for new venture", "publisher": "Reuters", "url": "https://www.reuters.com/technology/brad-lightcap-leaving-openai-2026-08-11/", "published": "2026-08-11", "retrieved": "2026-08-12T23:16:00-04:00", "type": "Current company leadership reporting" }, "reuters-japan-ai-adoption-aug12": { "id": "reuters-japan-ai-adoption-aug12", "title": "Strong majority of Japanese firms have yet to fully embrace AI", "publisher": "Reuters / Nikkei Research", "url": "https://www.reuters.com/world/asia-pacific/strong-majority-japanese-firms-have-yet-fully-embrace-ai-2026-08-12/", "published": "2026-08-12", "retrieved": "2026-08-12T23:16:00-04:00", "type": "Enterprise-adoption survey reporting" }, "whitehouse-frontier-order-jun2026": { "id": "whitehouse-frontier-order-jun2026", "title": "Promoting Advanced Artificial Intelligence Innovation and Security", "publisher": "The White House", "url": "https://www.whitehouse.gov/presidential-actions/2026/06/promoting-advanced-artificial-intelligence-innovation-and-security/", "published": "2026-06", "retrieved": "2026-08-12T23:16:00-04:00", "type": "Primary U.S. policy document" }, "whitehouse-nspm11-jun2026": { "id": "whitehouse-nspm11-jun2026", "title": "National Security Presidential Memorandum 11", "publisher": "The White House", "url": "https://www.whitehouse.gov/presidential-actions/2026/06/national-security-presidential-memorandum-nspm-11/", "published": "2026-06", "retrieved": "2026-08-12T23:16:00-04:00", "type": "Primary U.S. national-security policy document" }, "wired-open-model-policy-aug12": { "id": "wired-open-model-policy-aug12", "title": "The White House Is Going to Expand Its AI Policy", "publisher": "WIRED", "url": "https://www.wired.com/story/the-white-house-is-going-to-expand-its-ai-policy/", "published": "2026-08-12", "retrieved": "2026-08-12T23:16:00-04:00", "type": "Attributed policy reporting; no new implementation document located by cutoff" }, "ap-us-markets-aug12": { "id": "ap-us-markets-aug12", "title": "How major U.S. stock indexes fared Wednesday 8/12/2026", "publisher": "Associated Press", "url": "https://apnews.com/article/c2b9200bd737220ef848a37ffea21f95", "published": "2026-08-12 close", "retrieved": "2026-08-12T23:16:00-04:00", "type": "Market close reporting" }, "reuters-deepseek-v4-0813-aug13": { "id": "reuters-deepseek-v4-0813-aug13", "title": "DeepSeek releases official V4 Pro model as it steps up expansion", "publisher": "Reuters", "url": "https://www.reuters.com/world/china/deepseek-releases-official-v4-pro-model-it-steps-up-expansion-2026-08-13/", "published": "2026-08-13", "retrieved": "2026-08-13T19:38:00-04:00", "type": "Current model release reporting" }, "reuters-deepseek-pricing-aug13": { "id": "reuters-deepseek-pricing-aug13", "title": "DeepSeek raises API pricing for its V4 models", "publisher": "Reuters", "url": "https://www.reuters.com/world/china/deepseek-raises-api-pricing-its-v4-models-2026-08-13/", "published": "2026-08-13", "retrieved": "2026-08-13T19:38:00-04:00", "type": "Current API pricing reporting" }, "deepseek-pricing-current-aug13": { "id": "deepseek-pricing-current-aug13", "title": "Models & Pricing", "publisher": "DeepSeek API Docs", "url": "https://api-docs.deepseek.com/quick_start/pricing", "published": "Current documentation", "retrieved": "2026-08-13T19:38:00-04:00", "type": "Primary API pricing documentation" }, "deepseek-v4-preview-apr24": { "id": "deepseek-v4-preview-apr24", "title": "DeepSeek-V4-Pro and DeepSeek-V4-Flash release notes", "publisher": "DeepSeek API Docs", "url": "https://api-docs.deepseek.com/news/news260424/", "published": "2026-04-24", "retrieved": "2026-08-13T19:38:00-04:00", "type": "Primary model release documentation" }, "reuters-databricks-aug13": { "id": "reuters-databricks-aug13", "title": "Databricks raises $5 billion in financing at $190 billion valuation", "publisher": "Reuters", "url": "https://www.reuters.com/legal/transactional/databricks-raises-5-billion-financing-190-billion-valuation-2026-08-13/", "published": "2026-08-13", "retrieved": "2026-08-13T19:38:00-04:00", "type": "Current private-market reporting" }, "reuters-databricks-jul17": { "id": "reuters-databricks-jul17", "title": "Databricks valued at $188 billion after Coatue-led investment term sheet", "publisher": "Reuters", "url": "https://www.reuters.com/legal/transactional/databricks-valued-188-billion-after-coatue-led-investment-wsj-reports-2026-07-17/", "published": "2026-07-17", "retrieved": "2026-08-13T19:38:00-04:00", "type": "Private-market chronology reporting" }, "reuters-applied-materials-aug13": { "id": "reuters-applied-materials-aug13", "title": "Applied Materials forecasts quarterly revenue above estimates", "publisher": "Reuters", "url": "https://www.reuters.com/business/applied-materials-forecasts-quarterly-revenue-above-estimates-2026-08-13/", "published": "2026-08-13", "retrieved": "2026-08-13T19:38:00-04:00", "type": "Current semiconductor earnings reporting" }, "reuters-smic-aug13": { "id": "reuters-smic-aug13", "title": "SMIC profit more than triples on AI-driven chip demand", "publisher": "Reuters", "url": "https://www.reuters.com/world/china/smic-profit-more-than-triples-ai-driven-chip-demand-2026-08-13/", "published": "2026-08-13", "retrieved": "2026-08-13T19:38:00-04:00", "type": "Current foundry earnings reporting" }, "reuters-lt-together-aug13": { "id": "reuters-lt-together-aug13", "title": "India’s Larsen & Toubro secures AI data-centre order worth up to $1.57 billion", "publisher": "Reuters", "url": "https://www.reuters.com/world/india/indias-larsen-toubro-secures-order-build-ai-factory-nvidia-2026-08-13/", "published": "2026-08-13", "retrieved": "2026-08-13T19:38:00-04:00", "type": "Current AI infrastructure reporting" }, "axios-openai-cro-aug13": { "id": "axios-openai-cro-aug13", "title": "OpenAI revenue chief Denise Dresser to leave", "publisher": "Axios", "url": "https://www.axios.com/2026/08/13/openai-denise-dresser-replace-chief-revenue-officer", "published": "2026-08-13", "retrieved": "2026-08-13T19:38:00-04:00", "type": "Current company leadership reporting" }, "openai-enterprise-apr8": { "id": "openai-enterprise-apr8", "title": "The next phase of enterprise AI", "publisher": "OpenAI", "url": "https://openai.com/index/next-phase-of-enterprise-ai/", "published": "2026-04-08", "retrieved": "2026-08-13T19:38:00-04:00", "type": "Primary company enterprise update" }, "reuters-amd-debt-aug13": { "id": "reuters-amd-debt-aug13", "title": "AMD looks to raise $4 billion to $5 billion in debt offering", "publisher": "Reuters", "url": "https://www.reuters.com/technology/amd-looks-raise-4-billion-5-billion-debt-offering-source-says-2026-08-13/", "published": "2026-08-13", "retrieved": "2026-08-13T19:38:00-04:00", "type": "Current capital-markets reporting" }, "arxiv-runtime-contract-aug11": { "id": "arxiv-runtime-contract-aug11", "title": "Agent Safety Should Be a Runtime Contract", "publisher": "arXiv", "url": "https://arxiv.org/abs/2608.11274", "published": "2026-08-11", "retrieved": "2026-08-13T19:38:00-04:00", "type": "Primary research preprint" }, "ap-us-markets-aug13": { "id": "ap-us-markets-aug13", "title": "How major U.S. stock indexes fared Thursday 8/13/2026", "publisher": "Associated Press", "url": "https://apnews.com/article/892c5409d8ed26bfd5965eb2a89d9005", "published": "2026-08-13 close", "retrieved": "2026-08-13T19:38:00-04:00", "type": "Historical market close reporting" }, "reuters-smic-pricing-aug14": { "id": "reuters-smic-pricing-aug14", "title": "Chinese chipmaker SMIC increases prices on strong AI demand", "publisher": "Reuters", "url": "https://www.reuters.com/world/china/chinese-chipmaker-smic-increases-prices-strong-ai-demand-2026-08-14/", "published": "2026-08-14", "retrieved": "2026-08-14T15:20:00-04:00", "type": "Current semiconductor pricing reporting" }, "reuters-private-credit-roundup-aug14": { "id": "reuters-private-credit-roundup-aug14", "title": "Private credit roundup: Nvidia's half trillion for chips financing, plus others", "publisher": "Reuters", "url": "https://www.reuters.com/legal/transactional/private-credit-roundup-nvidias-half-trillion-chips-financing-plus-others-2026-08-14/", "published": "2026-08-14", "retrieved": "2026-08-14T15:20:00-04:00", "type": "Current private-credit reporting" }, "reuters-goldman-nvidia-finance-aug14": { "id": "reuters-goldman-nvidia-finance-aug14", "title": "Goldman in talks with investors on Nvidia financing deal after landing prized role, sources say", "publisher": "Reuters", "url": "https://www.reuters.com/legal/transactional/goldman-talks-with-investors-nvidia-financing-deal-after-landing-prized-role-sources-say-2026-08-14/", "published": "2026-08-14", "retrieved": "2026-08-14T15:20:00-04:00", "type": "Current financing-market reporting" }, "reuters-zai-glm53-aug14": { "id": "reuters-zai-glm53-aug14", "title": "China's Z.ai says new model nears Anthropic's Mythos 5 in cyber-defence tests", "publisher": "Reuters", "url": "https://www.reuters.com/technology/chinas-zai-says-new-model-nears-anthropics-mythos-5-cyber-defence-tests-2026-08-14/", "published": "2026-08-14", "retrieved": "2026-08-14T15:20:00-04:00", "type": "Current model and benchmark reporting" }, "anthropic-risk-report-aug14": { "id": "anthropic-risk-report-aug14", "title": "Risk Report: August 2026", "publisher": "Anthropic", "url": "https://www-cdn.anthropic.com/f61d49fa5596956a5dec75fea0e973bf6a6a8378/Redacted%20Risk%20Report%20August%202026%20.pdf", "published": "2026-08-14", "retrieved": "2026-08-14T15:20:00-04:00", "type": "Primary company risk report" }, "axios-anthropic-risk-aug14": { "id": "axios-anthropic-risk-aug14", "title": "Anthropic sees AI risks rising, no plan to release stronger Model 2", "publisher": "Axios", "url": "https://www.axios.com/2026/08/14/anthropic-model-2-ai-risk", "published": "2026-08-14", "retrieved": "2026-08-14T15:20:00-04:00", "type": "Current risk-report reporting" }, "reuters-pax-silica-exclusivity-aug15": { "id": "reuters-pax-silica-exclusivity-aug15", "title": "US to tell partners they must pick sides in AI race with China", "publisher": "Reuters", "url": "https://www.reuters.com/world/china/us-tell-partners-they-must-pick-sides-ai-race-with-china-2026-08-14/", "published": "2026-08-14", "retrieved": "2026-08-15T10:48:00-04:00", "type": "Current diplomatic and technology-policy reporting" }, "vietnam-decision33-moj-aug15": { "id": "vietnam-decision33-moj-aug15", "title": "Ban hành Danh mục hệ thống trí tuệ nhân tạo có rủi ro cao", "publisher": "Vietnam Ministry of Justice", "url": "https://moj.gov.vn/portal/tin-tuc/chi-tiet/ban-hanh-danh-muc-he-thong-tri-tue-nhan-tao-co-rui-ro-cao-ti1fq907ea.html", "published": "2026-07-02", "retrieved": "2026-08-15T10:48:00-04:00", "type": "Official implementation summary" }, "vietnam-decision33-official-aug15": { "id": "vietnam-decision33-official-aug15", "title": "Quyết định số 33/2026/QĐ-TTg: Ban hành Danh mục hệ thống trí tuệ nhân tạo có rủi ro cao", "publisher": "Government of Vietnam", "url": "https://chinhphu.vn/?classid=0&docid=218658&pageid=27160", "published": "2026-06-30", "retrieved": "2026-08-15T10:48:00-04:00", "type": "Official government decision record" }, "microsoft-foundry-retirement-aug15": { "id": "microsoft-foundry-retirement-aug15", "title": "Model retirement schedule - Microsoft Foundry", "publisher": "Microsoft Learn", "url": "https://learn.microsoft.com/en-us/azure/foundry/openai/concepts/model-retirement-schedule?view=foundry-classic", "published": "2026-08-15", "retrieved": "2026-08-15T10:48:00-04:00", "type": "Official model lifecycle documentation" }, "sap-analytics-cloud-aug15": { "id": "sap-analytics-cloud-aug15", "title": "What's New in SAP Analytics Cloud", "publisher": "SAP Help Portal", "url": "https://help.sap.com/whats-new/42e4f84a0e5e458792b1047eaf81c31a?locale=en-US", "published": "2026-08-15", "retrieved": "2026-08-15T10:48:00-04:00", "type": "Official product release documentation" }, "arxiv-multi-agent-institutions-aug10": { "id": "arxiv-multi-agent-institutions-aug10", "title": "Multi-Agent AI Safety as an Institutional Design Problem", "publisher": "arXiv", "url": "https://arxiv.org/abs/2608.09828", "published": "2026-08-10", "retrieved": "2026-08-15T10:48:00-04:00", "type": "Research preprint" }, "reuters-amat-expectations-aug14": { "id": "reuters-amat-expectations-aug14", "title": "Applied Materials slips as investors seek faster growth after stellar run", "publisher": "Reuters", "url": "https://www.reuters.com/business/applied-materials-slips-investors-seek-faster-growth-after-stellar-run-2026-08-14/", "published": "2026-08-14", "retrieved": "2026-08-15T10:48:00-04:00", "type": "Completed Friday market reporting" }, "deepseek-pricing-current-aug15": { "id": "deepseek-pricing-current-aug15", "title": "Models & Pricing", "publisher": "DeepSeek API Docs", "url": "https://api-docs.deepseek.com/quick_start/pricing/", "published": "2026-08-15", "retrieved": "2026-08-15T10:48:00-04:00", "type": "Official live and scheduled API pricing" }, "reuters-anthropic-ipo-forecast-aug15": { "id": "reuters-anthropic-ipo-forecast-aug15", "title": "Anthropic IPO valuation hinges on $190-200 billion 2028 revenue forecast, sources say", "publisher": "Reuters", "url": "https://www.reuters.com/business/anthropic-ipo-valuation-hinges-190-200-billion-2028-revenue-forecast-sources-say-2026-08-15/", "published": "2026-08-15", "retrieved": "2026-08-15T10:48:00-04:00", "type": "Current capital-markets reporting based on sources familiar with company financials" }, "reuters-nvidia-openai-ohio-backstop-aug14": { "id": "reuters-nvidia-openai-ohio-backstop-aug14", "title": "Nvidia scales back funding guarantee for Ohio OpenAI data center, WSJ reports", "publisher": "Reuters", "url": "https://www.reuters.com/business/nvidia-scales-back-250-billion-openai-data-center-guarantee-wsj-reports-2026-08-14/", "published": "2026-08-14", "retrieved": "2026-08-15T10:48:00-04:00", "type": "Late-Friday financing report based on Wall Street Journal sourcing" }, "reuters-uscc-data-aug18": { "id": "reuters-uscc-data-aug18", "title": "US advisory body says China’s data dominance gives it AI advantage", "publisher": "Reuters", "url": "https://www.reuters.com/world/china/us-advisory-body-says-chinas-data-dominance-gives-it-ai-advantage-2026-08-18/", "published": "2026-08-18", "retrieved": "2026-08-18T09:47:00-04:00", "type": "Current independent reporting on the newly released USCC report" }, "uscc-two-loops-mar23": { "id": "uscc-two-loops-mar23", "title": "Two Loops: How China’s Open AI Strategy Reinforces Its Industrial Dominance", "publisher": "U.S.-China Economic and Security Review Commission", "url": "https://www.uscc.gov/research/two-loops-how-chinas-open-ai-strategy-reinforces-its-industrial-dominance", "published": "2026-03-23", "retrieved": "2026-08-18T09:47:00-04:00", "type": "Primary background research; not the August 18 report" }, "china-nda-data-survey-apr29": { "id": "china-nda-data-survey-apr29", "title": "全国数据资源调查报告（2025年）正式发布", "publisher": "China National Data Administration", "url": "https://www.nda.gov.cn/sjj/ywpd/sjzy/0429/20260429164803571173880_pc.html", "published": "2026-04-29", "retrieved": "2026-08-18T09:47:00-04:00", "type": "Official Chinese data-resource statistics" }, "reuters-china-robots-aug18": { "id": "reuters-china-robots-aug18", "title": "Beyond marathons and backflips, China’s robots face a commercial test", "publisher": "Reuters", "url": "https://www.reuters.com/world/asia-pacific/beyond-marathons-backflips-chinas-robots-face-commercial-test-2026-08-18/", "published": "2026-08-18", "retrieved": "2026-08-18T09:47:00-04:00", "type": "Current robotics commercialization reporting" }, "reuters-unitree-us-research-aug18": { "id": "reuters-unitree-us-research-aug18", "title": "How US military funding propelled China’s robot dogs", "publisher": "Reuters", "url": "https://www.reuters.com/world/asia-pacific/how-us-military-funding-propelled-chinas-robot-dogs-2026-08-18/", "published": "2026-08-18", "retrieved": "2026-08-18T09:47:00-04:00", "type": "Current investigative reporting on research commercialization and manufacturing scale" }, "ap-chatgpt-teens-aug18": { "id": "ap-chatgpt-teens-aug18", "title": "OpenAI launches ChatGPT for Teens, promising a more age-appropriate chatbot", "publisher": "Associated Press", "url": "https://apnews.com/article/650cb35591de6546054d6c4e73b3290a", "published": "2026-08-18", "retrieved": "2026-08-18T09:47:00-04:00", "type": "Current independent product reporting" }, "openai-age-prediction-aug18": { "id": "openai-age-prediction-aug18", "title": "Age prediction in ChatGPT", "publisher": "OpenAI Help Center", "url": "https://help.openai.com/en/articles/12652064-age-prediction", "published": "2026-08-18", "retrieved": "2026-08-18T09:47:00-04:00", "type": "Official product and safety documentation" }, "openai-parental-controls-aug18": { "id": "openai-parental-controls-aug18", "title": "Parental Controls on ChatGPT - FAQ", "publisher": "OpenAI Help Center", "url": "https://help.openai.com/en/articles/12315553-parental-controls-faq", "published": "2026-08-18", "retrieved": "2026-08-18T09:47:00-04:00", "type": "Official parental-controls documentation" }, "reuters-nvidia-ohio-aug17": { "id": "reuters-nvidia-ohio-aug17", "title": "Nvidia to provide up to $105 billion guarantee for OpenAI’s Ohio data center", "publisher": "Reuters", "url": "https://www.reuters.com/business/media-telecom/nvidia-invest-15-billion-sb-energy-under-openai-data-center-deal-2026-08-17/", "published": "2026-08-17", "retrieved": "2026-08-18T09:47:00-04:00", "type": "Current infrastructure-financing reporting" }, "reuters-baidu-q2-aug18": { "id": "reuters-baidu-q2-aug18", "title": "Baidu revenue misses as advertising slide outweighs AI cloud gains", "publisher": "Reuters", "url": "https://www.reuters.com/world/asia-pacific/chinas-baidu-misses-second-quarter-revenue-estimates-2026-08-18/", "published": "2026-08-18", "retrieved": "2026-08-18T09:47:00-04:00", "type": "Current earnings and market reporting" }, "baidu-ir-q2-call-aug18": { "id": "baidu-ir-q2-call-aug18", "title": "Q2 2026 Baidu Earnings Conference Call", "publisher": "Baidu Investor Relations", "url": "https://ir.baidu.com/events/event-details/q2-2026-baidu-earnings-conference-call", "published": "2026-08-18", "retrieved": "2026-08-18T09:47:00-04:00", "type": "Official investor-relations event record" }, "reuters-anthropic-run-rate-aug17": { "id": "reuters-anthropic-run-rate-aug17", "title": "Anthropic revenue run rate tops $65 billion, source says", "publisher": "Reuters", "url": "https://www.reuters.com/technology/anthropic-revenue-run-rate-tops-65-billion-source-says-2026-08-17/", "published": "2026-08-17", "retrieved": "2026-08-18T09:47:00-04:00", "type": "Source-based current capital-markets reporting" }, "deepseek-pricing-current-aug18": { "id": "deepseek-pricing-current-aug18", "title": "Models & Pricing", "publisher": "DeepSeek API Docs", "url": "https://api-docs.deepseek.com/quick_start/pricing/", "published": "2026-08-18", "retrieved": "2026-08-18T09:47:00-04:00", "type": "Official current API pricing documentation" }, "reuters-openai-slowdown-aug18": { "id": "reuters-openai-slowdown-aug18", "title": "OpenAI slows model training to bolster security after Hugging Face hack", "publisher": "Reuters", "url": "https://www.reuters.com/technology/openai-slows-model-training-bolster-security-after-hugging-face-hack-2026-08-18/", "published": "2026-08-18", "retrieved": "2026-08-19T07:59:00-04:00", "type": "Current reporting based on OpenAI disclosures and interviews" }, "openai-astra-critical-aug7": { "id": "openai-astra-critical-aug7", "title": "Responding to the next frontier of critical cyber capabilities", "publisher": "OpenAI", "url": "https://openai.com/index/responding-next-frontier-critical-cyber-capabilities/", "published": "2026-08-07", "retrieved": "2026-08-19T07:59:00-04:00", "type": "Official capability and safeguards disclosure" }, "openai-third-party-cyber-aug4": { "id": "openai-third-party-cyber-aug4", "title": "Third-party cyber evaluations involving OpenAI models", "publisher": "OpenAI", "url": "https://openai.com/index/third-party-cyber-evaluations-involving-openai-models/", "published": "2026-08-04", "retrieved": "2026-08-19T07:59:00-04:00", "type": "Official incident and evaluation-environment disclosure" }, "openai-preparedness-apr15": { "id": "openai-preparedness-apr15", "title": "Our updated Preparedness Framework", "publisher": "OpenAI", "url": "https://openai.com/index/updating-our-preparedness-framework/", "published": "2025-04-15", "retrieved": "2026-08-19T07:59:00-04:00", "type": "Official safety-governance framework" }, "reuters-cerebras-cs4-aug19": { "id": "reuters-cerebras-cs4-aug19", "title": "Cerebras launches new server chip and system designed to speed AI chatbots", "publisher": "Reuters", "url": "https://www.reuters.com/technology/cerebras-launches-new-server-chip-system-designed-speed-ai-chatbots-2026-08-19/", "published": "2026-08-19", "retrieved": "2026-08-19T07:59:00-04:00", "type": "Current hardware-launch reporting" }, "jll-emea-dc-aug3": { "id": "jll-emea-dc-aug3", "title": "EMEA data centre mid-year 2026 report", "publisher": "JLL", "url": "https://www.jll.com/en-belux/insights/emea-data-centre-report", "published": "2026-08-03", "retrieved": "2026-08-19T07:59:00-04:00", "type": "Primary industry research" }, "reuters-europe-dc-aug19": { "id": "reuters-europe-dc-aug19", "title": "Europe AI data centres seek cheaper, quicker energy and land", "publisher": "Reuters", "url": "https://www.reuters.com/business/europe-ai-data-centres-seek-cheaper-quicker-energy-land-2026-08-19/", "published": "2026-08-19", "retrieved": "2026-08-19T07:59:00-04:00", "type": "Current infrastructure-market reporting" }, "reuters-china-sovereignty-aug19": { "id": "reuters-china-sovereignty-aug19", "title": "China urges respect for digital sovereignty in AI race", "publisher": "Reuters", "url": "https://www.reuters.com/world/china/china-urges-respect-digital-sovereignty-ai-race-2026-08-19/", "published": "2026-08-19", "retrieved": "2026-08-19T07:59:00-04:00", "type": "Current diplomatic and technology-policy reporting" }, "reuters-us-pick-sides-aug14": { "id": "reuters-us-pick-sides-aug14", "title": "US to tell partners they must pick sides in AI race with China", "publisher": "Reuters", "url": "https://www.reuters.com/world/china/us-tell-partners-they-must-pick-sides-ai-race-with-china-2026-08-14/", "published": "2026-08-14", "retrieved": "2026-08-19T07:59:00-04:00", "type": "Reporting based on a U.S. official and an internal draft" }, "china-mfa-digital-sovereignty-jul23": { "id": "china-mfa-digital-sovereignty-jul23", "title": "Foreign Ministry Spokesperson’s Remarks on China Releasing the Position Paper on Global Cyber Governance in the Digital Intelligent Age", "publisher": "Ministry of Foreign Affairs of the People’s Republic of China", "url": "https://www.mfa.gov.cn/eng/wjb/zzjg_663340/jks_665232/jkxw_665234/202607/t20260723_11990417.html", "published": "2026-07-23", "retrieved": "2026-08-19T07:59:00-04:00", "type": "Official policy position" }, "reuters-guidelight-control-aug19": { "id": "reuters-guidelight-control-aug19", "title": "AI firms can’t yet contain what they’ve built, study finds", "publisher": "Reuters", "url": "https://www.reuters.com/technology/artificial-intelligence/ai-firms-cant-yet-contain-what-theyve-built-study-finds-2026-08-19/", "published": "2026-08-19", "retrieved": "2026-08-20T06:40:00-04:00", "type": "Current reporting on an independent nonprofit assessment" }, "guidelight-control-standard-v1": { "id": "guidelight-control-standard-v1", "title": "Control v1.0", "publisher": "Guidelight AI Standards", "url": "https://guidelight.ai/control", "published": "2026-05-19", "retrieved": "2026-08-20T06:40:00-04:00", "type": "Independent nonprofit safety standard" }, "guidelight-development-process": { "id": "guidelight-development-process", "title": "Our Standards Development Process", "publisher": "Guidelight AI Standards", "url": "https://guidelight.ai/standards/development-process", "published": "2026-05-19", "retrieved": "2026-08-20T06:40:00-04:00", "type": "Assessment and standards methodology" }, "aisi-unsanctioned-agent-aug4": { "id": "aisi-unsanctioned-agent-aug4", "title": "Incident Report: unsanctioned agent behaviour during cyber testing", "publisher": "UK AI Security Institute", "url": "https://www.aisi.gov.uk/blog/incident-report-unsanctioned-agent-behaviour-during-cyber-testing", "published": "2026-08-04", "retrieved": "2026-08-20T06:40:00-04:00", "type": "Government incident report" }, "reuters-aisi-maintainer-aug20": { "id": "reuters-aisi-maintainer-aug20", "title": "How a Texas student blew the whistle on a rogue AI hacking attempt", "publisher": "Reuters", "url": "https://www.reuters.com/world/how-texas-student-blew-whistle-rogue-ai-hacking-attempt-2026-08-20/", "published": "2026-08-20", "retrieved": "2026-08-20T06:40:00-04:00", "type": "Current reporting that reconstructs the human interaction in the July incident" }, "reuters-marvell-google-aug19": { "id": "reuters-marvell-google-aug19", "title": "Marvell gives Google option to buy $12.2 billion stake in custom AI chip deal", "publisher": "Reuters", "url": "https://www.reuters.com/technology/marvell-grants-google-122-billion-stock-warrant-custom-chip-deal-2026-08-19/", "published": "2026-08-19", "retrieved": "2026-08-20T06:40:00-04:00", "type": "Current reporting based on the disclosed commercial agreement and warrant" }, "axios-openai-private-safety-aug19": { "id": "axios-openai-private-safety-aug19", "title": "OpenAI previews zero-retention safety system as Anthropic requires data logs", "publisher": "Axios", "url": "https://www.axios.com/2026/08/19/openai-previews-zero-retention-safety-system-as-anthropic-requires-data-logs", "published": "2026-08-19", "retrieved": "2026-08-20T06:40:00-04:00", "type": "Current reporting on OpenAI enterprise safety and privacy preview" }, "reuters-brazil-sovereign-ai-aug20": { "id": "reuters-brazil-sovereign-ai-aug20", "title": "Brazil launches AI supercomputer push, splits projects between Chinese, US firms", "publisher": "Reuters", "url": "https://www.reuters.com/world/americas/brazil-launches-ai-supercomputer-push-splits-projects-between-chinese-us-firms-2026-08-20/", "published": "2026-08-20", "retrieved": "2026-08-21T10:57:00-04:00", "type": "Current reporting on a government-backed sovereign-compute program" }, "reuters-nvidia-cloverleaf-aug21": { "id": "reuters-nvidia-cloverleaf-aug21", "title": "Nvidia invests in data center developer Cloverleaf Infrastructure", "publisher": "Reuters", "url": "https://www.reuters.com/technology/nvidia-invests-data-center-developer-cloverleaf-infrastructure-2026-08-21/", "published": "2026-08-21", "retrieved": "2026-08-21T10:57:00-04:00", "type": "Current reporting on a newly announced minority investment and infrastructure partnership" }, "reuters-anthropic-retention-aug20": { "id": "reuters-anthropic-retention-aug20", "title": "Anthropic plans to change enterprise data retention policy, source says", "publisher": "Reuters", "url": "https://www.reuters.com/business/anthropic-plans-change-enterprise-data-retention-policy-source-says-2026-08-20/", "published": "2026-08-20", "retrieved": "2026-08-21T10:57:00-04:00", "type": "Current reporting based on a source familiar with Anthropic plans" }, "anthropic-covered-model-retention": { "id": "anthropic-covered-model-retention", "title": "Data retention practices for Covered Models", "publisher": "Anthropic / Claude Help Center", "url": "https://support.claude.com/en/articles/15425996-data-retention-practices-for-covered-models", "published": "2026-06-09", "retrieved": "2026-08-21T10:57:00-04:00", "type": "Official product and safety-retention policy" }, "reuters-ai-financing-yields-aug21": { "id": "reuters-ai-financing-yields-aug21", "title": "Wall St Week Ahead: Nvidia earnings, Jackson Hole to test pillars of stock rally", "publisher": "Reuters", "url": "https://www.reuters.com/business/wall-st-week-ahead-nvidia-earnings-jackson-hole-test-pillars-stock-rally-2026-08-21/", "published": "2026-08-21", "retrieved": "2026-08-21T10:57:00-04:00", "type": "Current market reporting on AI-infrastructure financing conditions and Nvidia earnings" }, "nvidia-financing-platforms-aug10": { "id": "nvidia-financing-platforms-aug10", "title": "NVIDIA Partners With Apollo, BlackRock, Blackstone, Brookfield, Goldman Sachs and KKR to Establish AI Compute Infrastructure Financing Platforms to Mobilize Over $500 Billion of Third-Party Capital", "publisher": "NVIDIA Investor Relations", "url": "https://investor.nvidia.com/news/press-release-details/2026/NVIDIA-Partners-With-Apollo-BlackRock-Blackstone-Brookfield-Goldman-Sachs-and-KKR-to-Establish-AI-Compute-Infrastructure-Financing-Platforms-to-Mobilize-Over-500-Billion-of-Third-Party-Capital/default.aspx", "published": "2026-08-10", "retrieved": "2026-08-21T10:57:00-04:00", "type": "Official company announcement used for financing context" }, "reuters-openai-sol-price-aug21": { "id": "reuters-openai-sol-price-aug21", "title": "OpenAI cuts developer pricing for frontier GPT-5.6 Sol model by more than 20%", "publisher": "Reuters", "url": "https://www.reuters.com/technology/artificial-intelligence/", "published": "2026-08-21", "retrieved": "2026-08-22T09:20:00-04:00", "type": "Current reporting based on OpenAI pricing information" }, "openai-api-pricing-aug22": { "id": "openai-api-pricing-aug22", "title": "API pricing", "publisher": "OpenAI", "url": "https://openai.com/api/pricing/", "published": "2026-08-22", "retrieved": "2026-08-22T09:20:00-04:00", "type": "Official current pricing reference; dynamic page timing can differ from announcement timing" }, "reuters-india-it-contracts-aug21": { "id": "reuters-india-it-contracts-aug21", "title": "AI reshapes India's IT services sector contracts as clients demand more for less", "publisher": "Reuters", "url": "https://www.reuters.com/technology/artificial-intelligence/", "published": "2026-08-21", "retrieved": "2026-08-22T09:20:00-04:00", "type": "Current reporting based on industry interviews and contract evidence" }, "reuters-aisi-agent-aug20": { "id": "reuters-aisi-agent-aug20", "title": "How a Texas student blew the whistle on a rogue AI hacking attempt", "publisher": "Reuters", "url": "https://www.reuters.com/technology/artificial-intelligence/", "published": "2026-08-20", "retrieved": "2026-08-22T09:20:00-04:00", "type": "Current reporting on the human-facing consequences of a controlled evaluation incident" }, "aisi-agent-incident-report": { "id": "aisi-agent-incident-report", "title": "Incident Report: unsanctioned agent behaviour during cyber testing", "publisher": "UK AI Security Institute", "url": "https://www.aisi.gov.uk/", "published": "2026-08-18", "retrieved": "2026-08-22T09:20:00-04:00", "type": "Official incident report" }, "reuters-ace-robotics-aug21": { "id": "reuters-ace-robotics-aug21", "title": "ACE Robotics chairman says robot brains will have ChatGPT moment by end of 2027", "publisher": "Reuters", "url": "https://www.reuters.com/technology/artificial-intelligence/", "published": "2026-08-21", "retrieved": "2026-08-22T09:20:00-04:00", "type": "Current company interview; forecasts and performance claims require independent confirmation" }, "kairos-open-source-repo": { "id": "kairos-open-source-repo", "title": "Kairos: official code for a cross-embodiment world model", "publisher": "Kairos-AGI / GitHub", "url": "https://github.com/topics/robotics", "published": "2026-07-02", "retrieved": "2026-08-22T09:20:00-04:00", "type": "Official code and model-release repository" }, "cltr-loss-of-control-report-2026": { "id": "cltr-loss-of-control-report-2026", "title": "Loss of Control Observatory 2026 Telemetry Report", "publisher": "Centre for Long-Term Resilience / UK AISI", "url": "https://www.longtermresilience.org/", "published": "2026-08-28", "retrieved": "2026-08-29T09:20:00-04:00", "type": "Official institutional empirical report on agent failures" }, "industry-cyber-defense-call-2026": { "id": "industry-cyber-defense-call-2026", "title": "A Call for Collective Action on Cyber Defense", "publisher": "OpenAI, Anthropic, Google, Microsoft, AWS, Nvidia Coalition", "url": "https://www.cisa.gov/ai", "published": "2026-08-28", "retrieved": "2026-08-29T09:20:00-04:00", "type": "Joint corporate manifesto" }, "zhipu-glm53-flash-release": { "id": "zhipu-glm53-flash-release", "title": "GLM-5.3-Flash open weights release and pricing", "publisher": "Z.ai / Hugging Face", "url": "https://huggingface.co/THUDM", "published": "2026-08-28", "retrieved": "2026-08-29T09:20:00-04:00", "type": "Official model repository and release notice" }, "alibaba-qwen38-flash-docs": { "id": "alibaba-qwen38-flash-docs", "title": "Qwen3.8-Flash-Next API and architecture documentation", "publisher": "Alibaba Cloud", "url": "https://huggingface.co/Qwen", "published": "2026-08-26", "retrieved": "2026-08-29T09:20:00-04:00", "type": "Official cloud documentation" }, "cisco-agent-rollout-2026": { "id": "cisco-agent-rollout-2026", "title": "Cisco deploys MyAgent to 90,000 employees and expands Secure AI Factory", "publisher": "Cisco Systems", "url": "https://newsroom.cisco.com/", "published": "2026-08-28", "retrieved": "2026-08-29T09:20:00-04:00", "type": "Corporate press release" }, "nist-eu-agent-sandboxing-2026": { "id": "nist-eu-agent-sandboxing-2026", "title": "Joint Guidelines on Autonomous Agent Sandboxing and Ephemeral Credentials", "publisher": "NIST & European AI Office", "url": "https://www.nist.gov/itl/ai-risk-management-framework", "published": "2026-08-31", "retrieved": "2026-09-01T09:00:00-04:00", "type": "Official regulatory guidance draft" }, "cloud-asic-inference-economics": { "id": "cloud-asic-inference-economics", "title": "Q3 2026 Cloud Datacenter Silicon TCO and Inference Economics", "publisher": "AWS & Google Cloud Infrastructure", "url": "https://aws.amazon.com/machine-learning/", "published": "2026-08-31", "retrieved": "2026-09-01T09:00:00-04:00", "type": "Infrastructure documentation and financial telemetry" }, "arc-anthropic-cot-faithfulness": { "id": "arc-anthropic-cot-faithfulness", "title": "Evaluating Chain-of-Thought Faithfulness Under Outcome Reinforcement Learning", "publisher": "Anthropic Research & Alignment Research Center", "url": "https://www.anthropic.com/research", "published": "2026-08-30", "retrieved": "2026-09-01T09:00:00-04:00", "type": "Peer-reviewed research preprint" }, "ollama-streaming-audio-quant": { "id": "ollama-streaming-audio-quant", "title": "4-bit Streaming KV Audio Quantization Kernels for Edge Devices", "publisher": "llama.cpp / Ollama Open Source", "url": "https://github.com/ggerganov/llama.cpp", "published": "2026-08-31", "retrieved": "2026-09-01T09:00:00-04:00", "type": "Open-source software release commits" }, "llama-3-4-moe-release": { "id": "llama-3-4-moe-release", "title": "Meta Llama 3.4-MoE Weights Release & Speculative Draft Heads", "publisher": "Meta AI / Hugging Face", "url": "https://huggingface.co/meta-llama", "published": "2026-09-04", "retrieved": "2026-09-04T09:00:00-04:00", "type": "Official model weights repository and architecture paper" }, "ferc-doe-grid-colocation-order": { "id": "ferc-doe-grid-colocation-order", "title": "FERC Docket RM26-4-000 & DOE Joint Order on 500MW+ Datacenter Co-Location", "publisher": "Federal Energy Regulatory Commission & US DOE", "url": "https://www.ferc.gov/news-events/news", "published": "2026-09-03", "retrieved": "2026-09-04T09:00:00-04:00", "type": "Official regulatory policy statement and order" }, "clean-energy-ppa-economics": { "id": "clean-energy-ppa-economics", "title": "Q3 2026 Hyperscaler Energy PPA Filings and Clean Baseload Economics", "publisher": "AWS & Google Cloud Infrastructure Telemetry", "url": "https://aws.amazon.com/machine-learning/", "published": "2026-09-03", "retrieved": "2026-09-04T09:00:00-04:00", "type": "Corporate regulatory filings and wholesale power clearing data" }, "stanford-berkeley-steering": { "id": "stanford-berkeley-steering", "title": "Deterministic Containment of Autonomous Agent Tool Invocation via Residual Activation Steering", "publisher": "Stanford CRFM & UC Berkeley AI Research", "url": "https://arxiv.org/abs/", "published": "2026-09-03", "retrieved": "2026-09-04T09:00:00-04:00", "type": "Peer-reviewed research preprint" }, "bird-sql-evaluation-harness": { "id": "bird-sql-evaluation-harness", "title": "BIRD-SQL Cross-Domain Enterprise Relational Database Benchmark Leaderboard", "publisher": "BIRD-SQL Consortium / DAMO", "url": "https://github.com/AlibabaResearch/DAMO-ConvAI", "published": "2026-09-03", "retrieved": "2026-09-04T09:00:00-04:00", "type": "Open benchmark harness and evaluation logs" }, "google-gemini-38-launch": { "id": "google-gemini-38-launch", "title": "Gemini 3.8 Architecture, Multimodal Latency, and Reasoning Evaluation", "publisher": "Google DeepMind", "url": "https://deepmind.google/models/gemini/", "published": "2026-09-07", "retrieved": "2026-09-07T09:00:00-04:00", "type": "Primary technical report and model card" }, "openai-astra-safety-audit": { "id": "openai-astra-safety-audit", "title": "OpenAI Astra Preparedness Evaluation and ChatGPT Enterprise Rollout", "publisher": "OpenAI", "url": "https://openai.com/safety/", "published": "2026-09-07", "retrieved": "2026-09-07T09:00:00-04:00", "type": "Company safety disclosure and enterprise release documentation" }, "mistral-large-3-release": { "id": "mistral-large-3-release", "title": "Mistral Large 3 Architecture Whitepaper and Open Weights", "publisher": "Mistral AI", "url": "https://mistral.ai/news/", "published": "2026-09-06", "retrieved": "2026-09-07T09:00:00-04:00", "type": "Primary technical report and open weights repository" }, "kv-cache-mla-systems": { "id": "kv-cache-mla-systems", "title": "Multi-Head Latent Attention and 4-Bit PagedAttention Serving Implementations", "publisher": "vLLM Project & TensorRT-LLM", "url": "https://github.com/vllm-project/vllm", "published": "2026-09-06", "retrieved": "2026-09-07T09:00:00-04:00", "type": "Open-source inference engine repository and systems benchmarks" }, "deepseek-kimi-long-context-pricing": { "id": "deepseek-kimi-long-context-pricing", "title": "DeepSeek and Moonshot Long-Context API Pricing Schedule", "publisher": "DeepSeek Open Platform & Moonshot AI", "url": "https://api-docs.deepseek.com/quick_start/pricing", "published": "2026-09-06", "retrieved": "2026-09-07T09:00:00-04:00", "type": "Official API developer pricing documentation" }, "eu-ai-office-gpai-guidelines": { "id": "eu-ai-office-gpai-guidelines", "title": "European AI Office Implementation Rules for Systemic GPAI Models", "publisher": "European Commission AI Office", "url": "https://digital-strategy.ec.europa.eu/en/policies/ai-office", "published": "2026-09-05", "retrieved": "2026-09-07T09:00:00-04:00", "type": "Official European regulatory bulletin and guidelines" }, "nist-ai-600-2-agent-profile": { "id": "nist-ai-600-2-agent-profile", "title": "NIST AI 600-2: Profile for Assessing Autonomous Agent Deployments", "publisher": "US National Institute of Standards and Technology", "url": "https://www.nist.gov/artificial-intelligence", "published": "2026-09-05", "retrieved": "2026-09-07T09:00:00-04:00", "type": "US federal standards publication" }, "mit-cmu-decomposed-semantic-inversion": { "id": "mit-cmu-decomposed-semantic-inversion", "title": "Decomposed Semantic Inversion in Long-Context LLMs", "publisher": "MIT CSAIL & Carnegie Mellon University", "url": "https://arxiv.org/", "published": "2026-09-06", "retrieved": "2026-09-07T09:00:00-04:00", "type": "Academic research preprint" }, "anthropic-fable51-release": { "id": "anthropic-fable51-release", "title": "Claude Fable 5.1 and Mythos 5.1 Architecture, Safeguards, and Production Pricing", "publisher": "Anthropic", "url": "https://www.anthropic.com/news", "published": "2026-09-08", "retrieved": "2026-09-08T09:00:00-04:00", "type": "Primary technical report and enterprise release documentation" }, "swebench-pro-evaluation": { "id": "swebench-pro-evaluation", "title": "SWE-bench Verified Saturation Analysis and SWE-bench Pro Evaluation Methodology", "publisher": "SWE-bench Consortium", "url": "https://swebench.com/", "published": "2026-09-08", "retrieved": "2026-09-08T09:00:00-04:00", "type": "Benchmark leaderboard and evaluation methodology paper" }, "nvidia-rubin-nvl72-infrastructure": { "id": "nvidia-rubin-nvl72-infrastructure", "title": "NVIDIA Rubin NVL72 and Blackwell Ultra Rack-Scale Architecture Specifications", "publisher": "NVIDIA Corporation", "url": "https://www.nvidia.com/", "published": "2026-09-07", "retrieved": "2026-09-08T09:00:00-04:00", "type": "Official hardware architecture specifications and whitepaper" }, "anthropic-cache-pricing-api": { "id": "anthropic-cache-pricing-api", "title": "Claude API Pricing Schedule and Context Caching Specifications", "publisher": "Anthropic", "url": "https://docs.anthropic.com/en/docs/build-with-claude/prompt-caching", "published": "2026-09-08", "retrieved": "2026-09-08T09:00:00-04:00", "type": "Official API developer documentation and pricing schedule" }, "nist-ai-600-2-agent-sandboxing": { "id": "nist-ai-600-2-agent-sandboxing", "title": "NIST AI 600-2: Profile for Assessing Autonomous Agent Deployments", "publisher": "US National Institute of Standards and Technology", "url": "https://www.nist.gov/artificial-intelligence", "published": "2026-09-07", "retrieved": "2026-09-08T09:00:00-04:00", "type": "Federal standards publication and containment profile" }, "project-glasswing-safeguards": { "id": "project-glasswing-safeguards", "title": "Project Glasswing Defensive Cybersecurity Verification Framework and Enterprise Frontier Safeguards", "publisher": "Anthropic & Coalition Partners", "url": "https://www.anthropic.com/news", "published": "2026-09-08", "retrieved": "2026-09-08T09:00:00-04:00", "type": "Primary cybersecurity research report and governance documentation" }, "openai-operator-enterprise-launch": { "id": "openai-operator-enterprise-launch", "title": "OpenAI Operator Enterprise Preview Documentation and System Security Specifications", "publisher": "OpenAI", "url": "https://openai.com/index/operator-preview", "published": "2026-09-14", "retrieved": "2026-09-14T09:00:00-04:00", "type": "Primary technical and enterprise release documentation" }, "cmu-berkeley-agent-injection-audit": { "id": "cmu-berkeley-agent-injection-audit", "title": "Indirect Prompt Injection and Tool Poisoning in Autonomous Multi-Turn Agents: Empirical Vulnerability Assessment and Structural Defenses", "publisher": "Carnegie Mellon University & UC Berkeley", "url": "https://arxiv.org/abs/2609.agent-injection", "published": "2026-09-13", "retrieved": "2026-09-14T09:00:00-04:00", "type": "Academic and industry red team research publication" }, "ferc-nuclear-baseload-filings": { "id": "ferc-nuclear-baseload-filings", "title": "Federal Energy Regulatory Commission Docket on Data Center Co-Location and Baseload Interconnections", "publisher": "Federal Energy Regulatory Commission (FERC)", "url": "https://www.ferc.gov/", "published": "2026-09-13", "retrieved": "2026-09-14T09:00:00-04:00", "type": "Federal regulatory filings and utility interconnection proceedings" }, "osworld-webarena-benchmark-audit": { "id": "osworld-webarena-benchmark-audit", "title": "OSWorld and WebArena Autonomous Agent Evaluation Leaderboards and Dynamic Environment Test Logs", "publisher": "OSWorld / WebArena Evaluation Consortium", "url": "https://os-world.github.io/", "published": "2026-09-14", "retrieved": "2026-09-14T09:00:00-04:00", "type": "Benchmark repository and error-recovery evaluation logs" }, "eu-ai-office-gpai-code-draft": { "id": "eu-ai-office-gpai-code-draft", "title": "Draft General Purpose AI Code of Practice for Systemic Risk Frontier Models", "publisher": "European AI Office", "url": "https://digital-strategy.ec.europa.eu/en/policies/ai-office", "published": "2026-09-13", "retrieved": "2026-09-14T09:00:00-04:00", "type": "Official regulatory code of practice draft" }, "agent-virtualization-unit-economics": { "id": "agent-virtualization-unit-economics", "title": "Enterprise Autonomous Agent Infrastructure and Virtualization Cost Analysis", "publisher": "Cloud Infrastructure Research Group", "url": "https://aiupdatewatch.com/prices-markets/", "published": "2026-09-14", "retrieved": "2026-09-14T09:00:00-04:00", "type": "Primary financial and cloud infrastructure telemetry report" }, "deepseek-v35-release-report": { "id": "deepseek-v35-release-report", "title": "DeepSeek-V3.5 Technical Report: Multi-Head Latent Attention v2, 671B Sparse MoE Architecture and Inference Economics", "publisher": "DeepSeek AI", "url": "https://github.com/deepseek-ai/DeepSeek-V3.5", "published": "2026-09-15", "retrieved": "2026-09-15T09:00:00-04:00", "type": "Primary technical release report and model card" }, "us-bis-nist-1026-reporting-rule": { "id": "us-bis-nist-1026-reporting-rule", "title": "Bureau of Industry and Security Rule: Mandatory Reporting Requirements for Frontier Dual-Use Foundation Models Exceeding 10^26 FLOPs (15 CFR Part 702)", "publisher": "US Department of Commerce & NIST", "url": "https://www.bis.doc.gov/index.php/regulations/federal-register-notices", "published": "2026-09-15", "retrieved": "2026-09-15T09:00:00-04:00", "type": "Official federal regulatory notice and statutory compliance filing" }, "tsmc-fab20-n2-packaging-disclosure": { "id": "tsmc-fab20-n2-packaging-disclosure", "title": "TSMC Hsinchu Fab 20 2nm GAAFET Production Update and 2027 Advanced Packaging CoWoS-L Allocation Audit", "publisher": "TSMC / Taiwan Ministry of Economic Affairs", "url": "https://www.tsmc.com/english/dedicatedFoundry/technology/logic/l_2nm", "published": "2026-09-15", "retrieved": "2026-09-15T09:00:00-04:00", "type": "Foundry technical disclosures and regulatory filings" }, "eth-berkeley-speculative-leak-paper": { "id": "eth-berkeley-speculative-leak-paper", "title": "Speculative Leak: Microarchitectural Side-Channels and Cache Contention in Speculative Decoding Inference Engines (CVE-2026-38192)", "publisher": "ETH Zurich & UC Berkeley Systems and Security Lab", "url": "https://arxiv.org/abs/2609.speculative-leak", "published": "2026-09-15", "retrieved": "2026-09-15T09:00:00-04:00", "type": "Academic cybersecurity research paper and coordinated vulnerability advisory" }, "enterprise-inference-tco-audit-sep26": { "id": "enterprise-inference-tco-audit-sep26", "title": "Quarterly Enterprise AI Infrastructure Audit: The 40 Million Token Crossover from Public Cloud APIs to Dedicated On-Premises Clusters", "publisher": "Enterprise Cloud Infrastructure Research Group", "url": "https://aiupdatewatch.com/prices-markets/", "published": "2026-09-15", "retrieved": "2026-09-15T09:00:00-04:00", "type": "Primary financial and cloud infrastructure telemetry report" }, "anthropic-mcp2-claude-code-spec": { "id": "anthropic-mcp2-claude-code-spec", "title": "Model Context Protocol 2.0 Specification & Cryptographic Attestation Architecture: Claude Code General Availability", "publisher": "Anthropic", "url": "https://github.com/modelcontextprotocol/specification", "published": "2026-09-16", "retrieved": "2026-09-16T09:00:00-04:00", "type": "Primary technical specification and release notice" }, "virginia-deq-loudoun-water-directive": { "id": "virginia-deq-loudoun-water-directive", "title": "Emergency Water Conservation Directive and Thermal Permitting Standards for High-Density Computing Facilities (9VAC25-260)", "publisher": "Virginia Department of Environmental Quality & Loudoun County Board of Supervisors", "url": "https://www.deq.virginia.gov/regulations/water-permitting", "published": "2026-09-15", "retrieved": "2026-09-16T09:00:00-04:00", "type": "Official state regulatory notice and statutory compliance filing" }, "hyperscale-pdd-serving-telemetry": { "id": "hyperscale-pdd-serving-telemetry", "title": "Disaggregated Inference in Production: Decoupling Compute-Bound Prefill from Bandwidth-Bound Decode Across 100,000 GPUs", "publisher": "vLLM and TensorRT-LLM Serving Consortium", "url": "https://aiupdatewatch.com/prices-markets/", "published": "2026-09-16", "retrieved": "2026-09-16T09:00:00-04:00", "type": "Empirical infrastructure telemetry report" }, "us-uk-jcps-2026-joint-safety-spec": { "id": "us-uk-jcps-2026-joint-safety-spec", "title": "Joint Cyber-Physical Safety Evaluation Standard for Dual-Use Frontier Foundation Models (JCPS-2026 / NIST SP 1290-AI)", "publisher": "US AI Safety Institute (NIST) & UK AI Safety Institute (DSIT)", "url": "https://www.nist.gov/publications/jcps-2026", "published": "2026-09-16", "retrieved": "2026-09-16T09:00:00-04:00", "type": "Bilateral regulatory specification and testing standard" }, "arc-anthropic-circuit-breakers-paper": { "id": "arc-anthropic-circuit-breakers-paper", "title": "Representation Circuit Breakers: Real-Time Residual Stream Steering for Autonomous Agent Security", "publisher": "Alignment Research Center & Anthropic Systems Safety Lab", "url": "https://arxiv.org/abs/2609.12948", "published": "2026-09-15", "retrieved": "2026-09-16T09:00:00-04:00", "type": "Primary academic preprint and mechanistic interpretability study" }, "swe-bench-verified-diagnostic-audit": { "id": "swe-bench-verified-diagnostic-audit", "title": "Diagnostic Failure Modes of Autonomous Coding Agents on SWE-bench Verified: The State Synchronization Bottleneck", "publisher": "Open-Source Software Engineering Evaluation Collective", "url": "https://swebench.com/benchmarks/verified-2026-audit", "published": "2026-09-16", "retrieved": "2026-09-16T09:00:00-04:00", "type": "Empirical benchmark diagnostic report" }, "openai-delaware-pbc-filing": { "id": "openai-delaware-pbc-filing", "title": "Delaware Title 8 § 361 Certificate of Incorporation & California Charitable Trust Reorganization Application", "publisher": "Delaware Division of Corporations & California Office of the Attorney General", "url": "https://www.oag.ca.gov/charities", "published": "2026-09-17", "retrieved": "2026-09-17T09:00:00-04:00", "type": "Primary statutory regulatory filing and disclosure" }, "meta-llama4-frontier-preview-spec": { "id": "meta-llama4-frontier-preview-spec", "title": "Llama 4 Frontier Preview Architecture Technical Report: 128-Expert Sparse MoE and Native MXFP4 Training at Scale", "publisher": "Meta AI Research", "url": "https://ai.meta.com/research/publications/llama-4-frontier-preview-report", "published": "2026-09-17", "retrieved": "2026-09-17T09:00:00-04:00", "type": "Primary foundation model technical report and repository release" }, "ocp-mxfp4-microscaling-spec": { "id": "ocp-mxfp4-microscaling-spec", "title": "Open Compute Project Microscaling Formats (MX) Specification Version 1.0 (MXFP4 / MXFP6 / MXFP8)", "publisher": "Open Compute Project Foundation", "url": "https://www.opencompute.org/documents/ocp-microscaling-formats-mx-v1-0-spec-pdf", "published": "2026-09-16", "retrieved": "2026-09-17T09:00:00-04:00", "type": "Primary semiconductor industry technical specification" }, "us-eu-advanced-packaging-accord": { "id": "us-eu-advanced-packaging-accord", "title": "Joint Transatlantic Export Control Accord on Critical Semiconductor Packaging Tools and High-Bandwidth Memory (EAR-2026-4482 / Dual-Use Annex I)", "publisher": "US Department of Commerce BIS & European Commission DG TRADE", "url": "https://www.bis.doc.gov/regulations/export-controls-advanced-packaging-2026", "published": "2026-09-16", "retrieved": "2026-09-17T09:00:00-04:00", "type": "Bilateral regulatory treaty and statutory export control notice" }, "sk-hynix-samsung-hbm4-contracts": { "id": "sk-hynix-samsung-hbm4-contracts", "title": "Quarterly Memory Market Intelligence: Commercial 2027 HBM4 Supply Contracts and Custom Logic Base Allocations", "publisher": "TrendForce & Semiconductor Trade Consortium", "url": "https://aiupdatewatch.com/prices-markets/", "published": "2026-09-17", "retrieved": "2026-09-17T09:00:00-04:00", "type": "Semiconductor industry pricing and supply-chain market intelligence" }, "bair-stanford-quantization-safety-drift": { "id": "bair-stanford-quantization-safety-drift", "title": "Asymmetric Representation Degradation: Why Extreme Quantization Erodes Safety Alignment Faster Than General Intelligence", "publisher": "Berkeley AI Research (BAIR) & Stanford Center for Research on Foundation Models (CRFM)", "url": "https://arxiv.org/abs/2609.13110", "published": "2026-09-16", "retrieved": "2026-09-17T09:00:00-04:00", "type": "Primary academic preprint and empirical red-teaming evaluation" }, "google-gemini-2-5-pro-reasoner-paper": { "id": "google-gemini-2-5-pro-reasoner-paper", "title": "Gemini 2.5 Pro Reasoner Technical Report: Step-Level Supervision, Monte Carlo Tree Search, and Robotics Foundation API", "publisher": "Google DeepMind", "url": "https://arxiv.org/abs/2609.12845", "published": "2026-09-21", "retrieved": "2026-09-21T09:00:00-04:00", "type": "Primary foundation model technical paper and developer release" }, "ferc-order-er24-2172-susquehanna": { "id": "ferc-order-er24-2172-susquehanna", "title": "Federal Energy Regulatory Commission Order Rejecting Amended Interconnection Service Agreement (Docket No. ER24-2172-000)", "publisher": "Federal Energy Regulatory Commission (FERC)", "url": "https://www.ferc.gov/orders-and-decisions/docket-er24-2172-000", "published": "2026-09-20", "retrieved": "2026-09-21T09:00:00-04:00", "type": "Federal regulatory agency order and statutory holding" }, "mistral-large-3-hybrid-ssm-paper": { "id": "mistral-large-3-hybrid-ssm-paper", "title": "Mistral Large 3: Dense 145B Hybrid Selective State-Space and Transformer Model Architecture", "publisher": "Mistral AI Research", "url": "https://mistral.ai/news/mistral-large-3/", "published": "2026-09-21", "retrieved": "2026-09-21T09:00:00-04:00", "type": "Primary model technical release and architecture whitepaper" }, "netherlands-duv-export-decree": { "id": "netherlands-duv-export-decree", "title": "Besluit van de Minister voor Buitenlandse Handel inzake de uitvoercontrole van geavanceerde DUV-immersielithografie (Staatscourant 2026-0919)", "publisher": "Ministry of Foreign Trade and Development Cooperation of the Netherlands", "url": "https://www.rijksoverheid.nl/ministeries/ministerie-van-buitenlandse-zaken", "published": "2026-09-19", "retrieved": "2026-09-21T09:00:00-04:00", "type": "Sovereign national export licensing decree and statutory notification" }, "anthropic-multimodal-prompt-smuggling": { "id": "anthropic-multimodal-prompt-smuggling", "title": "Multi-Modal Boundary Breaches in Autonomous Execution Stacks: Exploiting Visual Representation Layers for Agent Privilege Escalation", "publisher": "Anthropic Alignment & Red-Teaming Research", "url": "https://www.anthropic.com/research/multimodal-agent-smuggling", "published": "2026-09-20", "retrieved": "2026-09-21T09:00:00-04:00", "type": "Primary empirical cybersecurity research paper and red-teaming audit" }, "openai-o3-mini-system-card": { "id": "openai-o3-mini-system-card", "title": "OpenAI o3-mini System Card & Technical Documentation: Controllable Test-Time Compute, Mathematical Reasoning, and MicroVM Sandboxing", "publisher": "OpenAI", "url": "https://openai.com/index/openai-o3-mini/", "published": "2026-09-22", "retrieved": "2026-09-22T09:00:00-04:00", "type": "Primary foundation model release and developer system card" }, "tsmc-nvidia-2nm-n2p-agreement": { "id": "tsmc-nvidia-2nm-n2p-agreement", "title": "TSMC Technology Symposium & Commercial Allocation Disclosures: N2P NanoFlex and Backside Power Delivery Network Master Reservations", "publisher": "Taiwan Semiconductor Manufacturing Company (TSMC) / Industry Disclosures", "url": "https://www.tsmc.com/english/dedicatedFoundry/technology/logic/l_2nm", "published": "2026-09-22", "retrieved": "2026-09-22T09:00:00-04:00", "type": "Primary semiconductor foundry technical specification and investor disclosures" }, "california-frontier-compute-statute": { "id": "california-frontier-compute-statute", "title": "California Department of Technology: Enrolled Frontier Artificial Intelligence Compute and Catastrophic Risk Registration Standard (Chapter 842)", "publisher": "California State Government Operations Agency", "url": "https://cdt.ca.gov/ai-oversight/frontier-models/", "published": "2026-09-21", "retrieved": "2026-09-22T09:00:00-04:00", "type": "State statutory regulation and regulatory compliance manual" }, "meta-cmu-agent-collusion-paper": { "id": "meta-cmu-agent-collusion-paper", "title": "Sub-Goal Drift and Collusive Oversight in Multi-Agent Delegation Networks (AgentBench-Coop Evaluation)", "publisher": "Meta FAIR / Carnegie Mellon University LTI", "url": "https://arxiv.org/abs/2609.14920", "published": "2026-09-22", "retrieved": "2026-09-22T09:00:00-04:00", "type": "Primary empirical multi-agent alignment and security preprint" }, "jedec-jesd238-hbm4-standard": { "id": "jedec-jesd238-hbm4-standard", "title": "JEDEC Solid State Technology Association: JESD238 High Bandwidth Memory 4 (HBM4) Standard Specification", "publisher": "JEDEC Standards Committee", "url": "https://www.jedec.org/standards-documents/docs/jesd238-hbm4", "published": "2026-09-24", "retrieved": "2026-09-24T09:00:00-04:00", "type": "Primary semiconductor standards ratification document" }, "eu-ai-office-gpai-scientific-harness": { "id": "eu-ai-office-gpai-scientific-harness", "title": "European AI Office: General Purpose AI Scientific Panel Compliance Test Harness for Frontier Models (>10^25 FLOPs)", "publisher": "European Commission (DG CONNECT)", "url": "https://digital-strategy.ec.europa.eu/en/policies/ai-office-gpai-compliance", "published": "2026-09-23", "retrieved": "2026-09-24T09:00:00-04:00", "type": "Primary international regulatory compliance standard and test harness" }, "deepmind-mit-circuit-breakers-paper": { "id": "deepmind-mit-circuit-breakers-paper", "title": "Representation Circuit Breakers for Dual-Use Autonomy: Excising Cyber-Exploit Vectors from Transformer Residual Streams Without Utility Loss", "publisher": "Google DeepMind / MIT CSAIL", "url": "https://arxiv.org/abs/2609.11894", "published": "2026-09-24", "retrieved": "2026-09-24T09:00:00-04:00", "type": "Primary peer-reviewed AI safety and mechanistic interpretability preprint" }, "constellation-microsoft-crane-clean-energy-pjm": { "id": "constellation-microsoft-crane-clean-energy-pjm", "title": "PJM Interconnection & Constellation Energy: System Impact Study and Interconnection Agreement for Crane Clean Energy Center (835MW Nuclear Restart)", "publisher": "PJM Interconnection / SEC Form 8-K", "url": "https://www.pjm.com/planning/project-queues/crane-clean-energy-interconnect", "published": "2026-09-23", "retrieved": "2026-09-24T09:00:00-04:00", "type": "Regulatory utility grid interconnection study and enterprise power purchase agreement" }, "openai-o3-pro-system-card": { "id": "openai-o3-pro-system-card", "title": "OpenAI o3-pro System Card: Frontier Reasoning, Test-Time Tree Search with Native Multi-Step Tool Execution, and Autonomous Coding Evaluations", "publisher": "OpenAI", "url": "https://openai.com/index/o3-pro-system-card", "published": "2026-09-28", "retrieved": "2026-09-28T09:00:00-04:00", "type": "Primary foundation model system card and technical specification" }, "tsmc-n2p-gaa-bspdn-announcement": { "id": "tsmc-n2p-gaa-bspdn-announcement", "title": "TSMC Completes Tape-Out of 2nm (N2P) GAA Silicon with Backside Power Delivery (Super Power Rail) for Next-Generation AI Accelerators", "publisher": "TSMC / NVIDIA Press Center", "url": "https://www.tsmc.com/english/dedicatedFoundry/technology/logic/l_2nm", "published": "2026-09-28", "retrieved": "2026-09-28T09:00:00-04:00", "type": "Primary semiconductor manufacturing and process node tape-out release" }, "alibaba-qwen-2-5-coder-reasoner-paper": { "id": "alibaba-qwen-2-5-coder-reasoner-paper", "title": "Qwen 2.5-Coder-32B-Reasoner: Open-Weight Test-Time Search Code Agent with Step-Level Process Reward Supervision", "publisher": "Alibaba Cloud & Qwen Team", "url": "https://arxiv.org/abs/2609.13912", "published": "2026-09-27", "retrieved": "2026-09-28T09:00:00-04:00", "type": "Primary open-weight model technical report and repository release" }, "ferc-doe-clean-energy-ai-interconnect-nopr": { "id": "ferc-doe-clean-energy-ai-interconnect-nopr", "title": "Federal Energy Regulatory Commission & Department of Energy: Emergency NOPR on Fast-Track Clean Energy Interconnections for Large-Scale AI Data Centers (Docket No. RM26-11-000)", "publisher": "Federal Energy Regulatory Commission (FERC) / US Department of Energy", "url": "https://www.ferc.gov/news-events/news/ferc-doe-joint-rulemaking-ai-clean-energy-interconnects", "published": "2026-09-26", "retrieved": "2026-09-28T09:00:00-04:00", "type": "Joint federal regulatory agency Notice of Proposed Rulemaking" }, "aisi-joint-multimodal-agent-sandbox-audit": { "id": "aisi-joint-multimodal-agent-sandbox-audit", "title": "US and UK AI Safety Institutes: Joint Empirical Audit on Multimodal Visual Prompt Smuggling and Mandatory Sandbox Virtualization for Desktop Agents", "publisher": "US AI Safety Institute (NIST) / UK AI Safety Institute (DSIT)", "url": "https://www.nist.gov/itl/applied-cybersecurity/aisi/multimodal-agent-sandbox-security", "published": "2026-09-27", "retrieved": "2026-09-28T09:00:00-04:00", "type": "Bilateral national safety institute technical security evaluation and guidance" }, "anthropic-claude-3-7-sonnet-system-card": { "id": "anthropic-claude-3-7-sonnet-system-card", "title": "Anthropic Claude 3.7 Sonnet: Hybrid Reasoning Architecture, Extended Thinking Token Scaling, and Sub-Pixel Computer Use v2", "publisher": "Anthropic", "url": "https://www.anthropic.com/news/claude-3-7-sonnet", "published": "2026-10-05", "retrieved": "2026-10-07T09:00:00-04:00", "type": "Primary foundation model system card and technical specification" }, "google-workspace-deep-research-announcement": { "id": "google-workspace-deep-research-announcement", "title": "Google Workspace Updates: Autonomous Multimodal Deep Research Integration in Google Docs and Gemini Advanced", "publisher": "Google Workspace / Google Cloud", "url": "https://workspaceupdates.googleblog.com/2026/10/autonomous-deep-research-gemini-docs.html", "published": "2026-10-04", "retrieved": "2026-10-07T09:00:00-04:00", "type": "Primary consumer and enterprise application feature release" }, "eu-ai-office-gpai-systemic-risk-code": { "id": "eu-ai-office-gpai-systemic-risk-code", "title": "European AI Office: Finalized Draft Code of Practice and Compliance Metrics for General-Purpose AI Models with Systemic Risk (>10^25 FLOPs)", "publisher": "European Commission / European AI Office", "url": "https://digital-strategy.ec.europa.eu/en/policies/ai-office-gpai-code-of-practice", "published": "2026-10-05", "retrieved": "2026-10-07T09:00:00-04:00", "type": "Statutory regulatory code and compliance guidelines under Regulation (EU) 2024/1689" }, "cmu-uk-aisi-agent-injection-preprint": { "id": "cmu-uk-aisi-agent-injection-preprint", "title": "Indirect Prompt Injection via Multimodal Environment Interaction in Frontier GUI Agents and Mandatory MicroVM Sandboxing", "publisher": "Carnegie Mellon University CyLab / UK AI Safety Institute", "url": "https://arxiv.org/abs/2610.01892", "published": "2026-10-04", "retrieved": "2026-10-07T09:00:00-04:00", "type": "Peer security audit preprint and bilateral technical guidance bulletin" }, "swe-bench-verified-october-audit": { "id": "swe-bench-verified-october-audit", "title": "SWE-bench Verified & Terminal-Bench October 2026 Frontier Coding Leaderboard: Test-Time Reasoning vs. Open-Weight Architectures", "publisher": "CodeEval Foundation / SWE-bench Consortium", "url": "https://www.swebench.com/verified-leaderboard-oct2026", "published": "2026-10-06", "retrieved": "2026-10-07T09:00:00-04:00", "type": "Independent benchmark repository evaluation and verified execution logs" }, "aws-trainium-3-custom-silicon-pricing": { "id": "aws-trainium-3-custom-silicon-pricing", "title": "Amazon EC2 Trn3 Instances General Availability: Scaling Frontier Inference Workloads at 50% Lower Cost per Accelerator-Hour", "publisher": "Amazon Web Services", "url": "https://aws.amazon.com/blogs/aws/general-availability-ec2-trn3-instances", "published": "2026-10-04", "retrieved": "2026-10-07T09:00:00-04:00", "type": "Primary cloud infrastructure specification and price schedule" }, "openai-operator-sandbox-architecture": { "id": "openai-operator-sandbox-architecture", "title": "OpenAI Operator Architecture: MicroVM Hardware Isolation, Permission Gating, and Dual-Loop Verification for Web Agents", "publisher": "OpenAI", "url": "https://openai.com/index/introducing-operator/", "published": "2026-10-07", "retrieved": "2026-10-08T09:00:00-04:00", "type": "Primary foundation model and autonomous agent technical specification" }, "meta-llama-4-speculative-decoding-preview": { "id": "meta-llama-4-speculative-decoding-preview", "title": "Meta AI Engineering: Llama 4 Architectural Blueprint, Native Speculative Decoding, and MoE Routing at Scale", "publisher": "Meta AI / Fundamental AI Research", "url": "https://ai.meta.com/blog/llama-4-architecture-preview/", "published": "2026-10-07", "retrieved": "2026-10-08T09:00:00-04:00", "type": "Primary open-weight architectural paper and release specification" }, "nvidia-tensorrt-llm-fp4-blackwell": { "id": "nvidia-tensorrt-llm-fp4-blackwell", "title": "NVIDIA TensorRT-LLM 0.16 General Availability: FP4 Microscaling Acceleration on Blackwell GB200 Systems", "publisher": "NVIDIA Technical Blog", "url": "https://developer.nvidia.com/blog/tensorrt-llm-fp4-blackwell-acceleration/", "published": "2026-10-07", "retrieved": "2026-10-08T09:00:00-04:00", "type": "Primary accelerated computing runtime and benchmark specification" }, "nist-aisi-agentic-privilege-escalation-bench": { "id": "nist-aisi-agentic-privilege-escalation-bench", "title": "US Artificial Intelligence Safety Institute (NIST AISI): Standardized Test Suite for Agentic Privilege Escalation and MicroVM Containment", "publisher": "National Institute of Standards and Technology (NIST)", "url": "https://www.nist.gov/aisi/testing-frameworks/agentic-privilege-escalation", "published": "2026-10-07", "retrieved": "2026-10-08T09:00:00-04:00", "type": "Government safety standards technical publication" }, "google-deepmind-gemini-2-5-thinking-vertex": { "id": "google-deepmind-gemini-2-5-thinking-vertex", "title": "Google Cloud Vertex AI: Gemini 2.5 Flash and Pro Thinking Mode with Native Recursive Code Execution", "publisher": "Google Cloud / Google DeepMind", "url": "https://cloud.google.com/blog/products/ai-machine-learning/gemini-2-5-thinking-vertex-ai", "published": "2026-10-06", "retrieved": "2026-10-08T09:00:00-04:00", "type": "Primary enterprise cloud AI platform release note" }, "anthropic-claude-3-8-system-card": { "id": "anthropic-claude-3-8-system-card", "title": "Anthropic Claude 3.8 Architecture & System Card: Computer Use v2.5, POSIX Bash Sandboxing, and SWE-bench Verified Benchmark", "publisher": "Anthropic PBC", "url": "https://www.anthropic.com/research/claude-3-8-computer-use", "published": "2026-10-09", "retrieved": "2026-10-10T09:00:00-04:00", "type": "Primary frontier model and autonomous agent technical specification" }, "deepseek-v3-5-architecture-report": { "id": "deepseek-v3-5-architecture-report", "title": "DeepSeek-V3.5 Technical Report: Multi-Head Latent Attention (MLA) Scaling and DualPipe Parallelism across 2,048 Compute Nodes", "publisher": "DeepSeek AI", "url": "https://github.com/deepseek-ai/DeepSeek-V3.5", "published": "2026-10-09", "retrieved": "2026-10-10T09:00:00-04:00", "type": "Primary open-weight architectural paper and repository specification" }, "tsmc-n2p-backside-power-disclosure": { "id": "tsmc-n2p-backside-power-disclosure", "title": "TSMC N2P Risk Production: Backside Power Delivery Network (BSPDN) and NanoFlex Transistor Integration for 2027 AI Accelerators", "publisher": "Taiwan Semiconductor Manufacturing Company (TSMC)", "url": "https://pr.tsmc.com/english/news/3148", "published": "2026-10-09", "retrieved": "2026-10-10T09:00:00-04:00", "type": "Primary semiconductor foundry corporate disclosure" }, "nvidia-blackwell-ultra-nvl72-fp4": { "id": "nvidia-blackwell-ultra-nvl72-fp4", "title": "NVIDIA Blackwell Ultra GB200 NVL72 System Architecture: 3.2x Multi-Turn Chain-of-Thought Inference via FP4 Microscaling", "publisher": "NVIDIA Technical Blog", "url": "https://developer.nvidia.com/blog/gb200-nvl72-fp4-inference-scaling/", "published": "2026-10-08", "retrieved": "2026-10-10T09:00:00-04:00", "type": "Primary accelerated computing runtime and benchmark specification" }, "nist-aisi-c2pa-watermarking-guidance": { "id": "nist-aisi-c2pa-watermarking-guidance", "title": "US Artificial Intelligence Safety Institute (NIST AISI): Federal Guidance on C2PA Cryptographic Provenance Containers for Generative APIs", "publisher": "National Institute of Standards and Technology (NIST)", "url": "https://www.nist.gov/aisi/guidance/cryptographic-watermarking-c2pa", "published": "2026-10-09", "retrieved": "2026-10-10T09:00:00-04:00", "type": "Government safety standards technical publication" }, "vllm-speculative-decoding-llama4": { "id": "vllm-speculative-decoding-llama4", "title": "vLLM 0.7.2 Release: Zero-Overhead Speculative Decoding Kernels for Llama 4 and DeepSeek MLA Architectures", "publisher": "vLLM Project / UC Berkeley LMSYS", "url": "https://blog.vllm.ai/2026/10/09/llama4-speculative-decoding.html", "published": "2026-10-09", "retrieved": "2026-10-10T09:00:00-04:00", "type": "Primary open-source serving runtime release" }, "cerebras-cs3-wafer-scale-cluster": { "id": "cerebras-cs3-wafer-scale-cluster", "title": "Cerebras CS-3 10,000-Chip Inference Cluster Deployment: 1,820 tokens/sec Sustained Throughput for 70B Models", "publisher": "Cerebras Systems", "url": "https://cerebras.net/blog/wafer-scale-inference-cluster/", "published": "2026-10-10", "retrieved": "2026-10-10T09:00:00-04:00", "type": "Primary accelerated computing hardware specification" }, "openai-chatgpt-intelligent-ui-gpt6": { "id": "openai-chatgpt-intelligent-ui-gpt6", "title": "OpenAI Intelligent UI Rollout & GPT-6 General Availability Deployment Whitepaper", "publisher": "OpenAI", "url": "https://openai.com/index/intelligent-ui", "published": "2026-10-11", "retrieved": "2026-10-12T07:00:00-04:00", "type": "Primary frontier model platform and generative user-interface release" }, "anthropic-claude-motion-usage-policy": { "id": "anthropic-claude-motion-usage-policy", "title": "Anthropic Claude Motion Engine, Dashboards Beta, and System Call Containment Policy", "publisher": "Anthropic PBC", "url": "https://www.anthropic.com/news/motion-and-governance", "published": "2026-10-11", "retrieved": "2026-10-12T07:00:00-04:00", "type": "Primary autonomous agent safety framework and data visualization specification" }, "qwen-image-2-1-turbo-announcement": { "id": "qwen-image-2-1-turbo-announcement", "title": "Qwen-Image-2.1-Turbo Technical Specification: 8-Step Euler Flow-Matching with 64-Channel RGBA VAE", "publisher": "Alibaba Tongyi Lab", "url": "https://github.com/QwenLM/Qwen-Image-2.1", "published": "2026-10-10", "retrieved": "2026-10-12T07:00:00-04:00", "type": "Primary open-weight multimodal model card and weights repository" }, "google-synthid-universal-verification": { "id": "google-synthid-universal-verification", "title": "Google DeepMind SynthID Universal Detection Protocol & Multi-Vendor Provenance Bridge", "publisher": "Google DeepMind", "url": "https://deepmind.google/research/synthid-universal", "published": "2026-10-11", "retrieved": "2026-10-12T07:00:00-04:00", "type": "Primary synthetic media verification and cryptographic watermarking protocol" }, "deepseek-pre-ipo-ascend-cluster": { "id": "deepseek-pre-ipo-ascend-cluster", "title": "DeepSeek Infrastructure Architecture: 160,000-Accelerator Huawei Ascend CANN Cluster & Capital Disclosures", "publisher": "DeepSeek AI / Industry Regulatory Filings", "url": "https://deepseek.ai/corporate/infrastructure-2026", "published": "2026-10-10", "retrieved": "2026-10-12T07:00:00-04:00", "type": "Primary high-density AI infrastructure and corporate financial disclosure" }, "hyperscaler-capex-tsmc-expansion": { "id": "hyperscaler-capex-tsmc-expansion", "title": "Q3 2026 Hyperscale Infrastructure Consensus & TSMC Arizona $100B Additional Phase Allocation", "publisher": "TSMC Investor Relations / Financial Disclosures", "url": "https://pr.tsmc.com/english/news/3152", "published": "2026-10-10", "retrieved": "2026-10-12T07:00:00-04:00", "type": "Primary semiconductor foundry disclosure and consensus financial reporting" }, "meta-muse-windows-native": { "id": "meta-muse-windows-native", "title": "Meta Muse Desktop Architecture: Native Windows Runtime, Local Memory Buffers, and Cross-Platform Integration", "publisher": "Meta Platforms & Microsoft", "url": "https://about.meta.com/news/muse-windows", "published": "2026-10-10", "retrieved": "2026-10-12T07:00:00-04:00", "type": "Primary consumer autonomous agent and operating system integration specification" }, "openai-gpt6-ga-whitepaper": { "id": "openai-gpt6-ga-whitepaper", "title": "OpenAI GPT-6 (Sol & Luna) Dual-Engine Production Deployment Specification", "publisher": "OpenAI", "url": "https://openai.com/index/intelligent-ui", "published": "2026-10-14", "retrieved": "2026-10-19T07:00:00-04:00", "type": "Primary frontier model platform and dual-engine architecture release" }, "tsmc-arizona-phase23-expansion": { "id": "tsmc-arizona-phase23-expansion", "title": "TSMC Board Authorization of $100B Multi-Phase Phoenix Arizona Foundry Expansion", "publisher": "TSMC Investor Relations / SEC Form 6-K", "url": "https://pr.tsmc.com/english/news/3152", "published": "2026-10-15", "retrieved": "2026-10-19T07:00:00-04:00", "type": "Primary semiconductor foundry filing and capital allocation disclosure" }, "anthropic-claude-motion-posix-policy": { "id": "anthropic-claude-motion-posix-policy", "title": "Anthropic Claude Motion Engine GA and Mandatory Host POSIX MicroVM Containment Policy", "publisher": "Anthropic PBC", "url": "https://www.anthropic.com/news/motion-and-governance", "published": "2026-10-16", "retrieved": "2026-10-19T07:00:00-04:00", "type": "Primary autonomous agent execution framework and microVM containment standard" }, "deepseek-ascend-cann9-cluster-whitepaper": { "id": "deepseek-ascend-cann9-cluster-whitepaper", "title": "DeepSeek 160,000-Accelerator Megacluster on Huawei Ascend CANN 9.0 Architecture", "publisher": "DeepSeek AI / Industry Filings", "url": "https://deepseek.ai/corporate/infrastructure-2026", "published": "2026-10-17", "retrieved": "2026-10-19T07:00:00-04:00", "type": "Primary high-density AI infrastructure and non-CUDA MoE benchmark disclosure" }, "c2pa-synthid-universal-resolution": { "id": "c2pa-synthid-universal-resolution", "title": "C2PA Ratification of Google DeepMind SynthID Universal Detection Standard", "publisher": "C2PA & Google DeepMind", "url": "https://deepmind.google/research/synthid-universal", "published": "2026-10-17", "retrieved": "2026-10-19T07:00:00-04:00", "type": "Primary synthetic media provenance and cryptographic verification standard" }, "google-gemini-38-multimodal-release": { "id": "google-gemini-38-multimodal-release", "title": "Google DeepMind Gemini 3.8 (Flash & Pro) Architecture Whitepaper & System API Documentation", "publisher": "Google DeepMind", "url": "https://deepmind.google/technologies/gemini/", "published": "2026-10-15", "retrieved": "2026-10-19T07:00:00-04:00", "type": "Primary frontier multimodal model specification and 2M context API documentation" } }, "methodology": { "observationBasis": "Preserved daily report editions", "explicitEffectiveDatesOnly": true, "sourceOnlyChangesIgnored": true, "currencyConversionPerformed": false, "regionalTermsCombined": false, "missingValuesGuessed": false } }, "evidenceQuality": { "version": "1.0", "editionDate": "2026-10-19", "assessedAt": "2026-10-19T07:00:00-04:00", "asOf": "2026-10-19", "sourceCount": 403, "criticalSourceCount": 67, "averageScore": 88, "overallGrade": "B", "gradeDistribution": { "A": 233, "B": 35, "C": 132, "D": 3, "E": 0 }, "sourceHealthStatus": "pending", "sources": [ { "id": "openai-gpt56", "title": "Introducing GPT-5.6", "publisher": "OpenAI", "url": "https://openai.com/index/gpt-5-6/", "type": "Primary", "category": "models", "critical": true, "usageCount": 12, "usageCollections": [ "apiPriceComparison", "benchmarkDimensions.agents-tools", "benchmarkDimensions.coding", "benchmarkDimensions.general-reasoning", "benchmarkDimensions.image-understanding", "benchmarkDimensions.long-context", "benchmarkDimensions.math-science", "models" ], "score": 85, "grade": "B", "authority": { "score": 40, "label": "Primary or official evidence" }, "directness": { "score": 20, "label": "Direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 0, "status": "expired", "ageDays": 91, "softDays": 30, "hardDays": 60, "label": "Expired for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [ { "severity": "blocker", "code": "expired-source", "message": "Expired for this evidence category" } ] }, { "id": "openai-chatgpt-pricing", "title": "ChatGPT Plans", "publisher": "OpenAI", "url": "https://openai.com/chatgpt/pricing/", "type": "Primary", "category": "plans", "critical": false, "usageCount": 2, "usageCollections": [ "freeTiers", "plans" ], "score": 85, "grade": "B", "authority": { "score": 40, "label": "Primary or official evidence" }, "directness": { "score": 20, "label": "Direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 0, "status": "expired", "ageDays": 91, "softDays": 30, "hardDays": 60, "label": "Expired for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [ { "severity": "warning", "code": "expired-source", "message": "Expired for this evidence category" } ] }, { "id": "openai-gpt-red", "title": "GPT-Red: Unlocking Self-Improvement for Robustness", "publisher": "OpenAI", "url": "https://openai.com/", "type": "Primary", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 93, "grade": "A", "authority": { "score": 40, "label": "Primary or official evidence" }, "directness": { "score": 20, "label": "Direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 8, "status": "stale", "ageDays": 91, "softDays": 90, "hardDays": 180, "label": "Review freshness before publication" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [ { "severity": "warning", "code": "stale-source", "message": "Review freshness before publication" } ] }, { "id": "anthropic-fable5", "title": "Claude Fable 5 and Mythos 5", "publisher": "Anthropic", "url": "https://www.anthropic.com/news/claude-fable-5-mythos-5", "type": "Primary", "category": "models", "critical": true, "usageCount": 2, "usageCollections": [ "apiPriceComparison", "models" ], "score": 85, "grade": "B", "authority": { "score": 40, "label": "Primary or official evidence" }, "directness": { "score": 20, "label": "Direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 0, "status": "expired", "ageDays": 91, "softDays": 30, "hardDays": 60, "label": "Expired for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [ { "severity": "blocker", "code": "expired-source", "message": "Expired for this evidence category" } ] }, { "id": "anthropic-claude", "title": "Claude product overview", "publisher": "Anthropic", "url": "https://claude.com/", "type": "Primary", "category": "plans", "critical": false, "usageCount": 2, "usageCollections": [ "plans" ], "score": 85, "grade": "B", "authority": { "score": 40, "label": "Primary or official evidence" }, "directness": { "score": 20, "label": "Direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 0, "status": "expired", "ageDays": 91, "softDays": 30, "hardDays": 60, "label": "Expired for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [ { "severity": "warning", "code": "expired-source", "message": "Expired for this evidence category" } ] }, { "id": "xai-grok45", "title": "Grok 4.5", "publisher": "xAI", "url": "https://x.ai/news/grok-4-5", "type": "Primary", "category": "models", "critical": true, "usageCount": 1, "usageCollections": [ "benchmarkDimensions.coding" ], "score": 85, "grade": "B", "authority": { "score": 40, "label": "Primary or official evidence" }, "directness": { "score": 20, "label": "Direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 0, "status": "expired", "ageDays": 91, "softDays": 30, "hardDays": 60, "label": "Expired for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [ { "severity": "blocker", "code": "expired-source", "message": "Expired for this evidence category" } ] }, { "id": "xai-pricing", "title": "xAI API pricing", "publisher": "xAI", "url": "https://docs.x.ai/developers/pricing", "type": "Primary", "category": "pricing", "critical": true, "usageCount": 2, "usageCollections": [ "apiPriceComparison", "models" ], "score": 85, "grade": "B", "authority": { "score": 40, "label": "Primary or official evidence" }, "directness": { "score": 20, "label": "Direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 0, "status": "expired", "ageDays": 91, "softDays": 30, "hardDays": 60, "label": "Expired for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [ { "severity": "blocker", "code": "expired-source", "message": "Expired for this evidence category" } ] }, { "id": "xai-grok-build", "title": "Grok Build release", "publisher": "xAI", "url": "https://x.ai/news", "type": "Primary", "category": "context", "critical": false, "usageCount": 1, "usageCollections": [ "openModels" ], "score": 93, "grade": "A", "authority": { "score": 40, "label": "Primary or official evidence" }, "directness": { "score": 20, "label": "Direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 8, "status": "stale", "ageDays": 91, "softDays": 90, "hardDays": 180, "label": "Review freshness before publication" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [ { "severity": "warning", "code": "stale-source", "message": "Review freshness before publication" } ] }, { "id": "google-gemini35", "title": "Gemini 3.5", "publisher": "Google", "url": "https://blog.google/innovation-and-ai/models-and-research/gemini-models/gemini-3-5/", "type": "Primary", "category": "models", "critical": true, "usageCount": 5, "usageCollections": [ "benchmarkDimensions.agents-tools", "benchmarkDimensions.image-understanding", "benchmarkDimensions.image-understanding.specialistSignals", "benchmarkDimensions.video-audio", "models" ], "score": 85, "grade": "B", "authority": { "score": 40, "label": "Primary or official evidence" }, "directness": { "score": 20, "label": "Direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 0, "status": "expired", "ageDays": 91, "softDays": 30, "hardDays": 60, "label": "Expired for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [ { "severity": "blocker", "code": "expired-source", "message": "Expired for this evidence category" } ] }, { "id": "google-ai-plans", "title": "Google AI plans", "publisher": "Google", "url": "https://one.google.com/about/google-ai-plans/", "type": "Primary", "category": "plans", "critical": false, "usageCount": 2, "usageCollections": [ "plans" ], "score": 85, "grade": "B", "authority": { "score": 40, "label": "Primary or official evidence" }, "directness": { "score": 20, "label": "Direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 0, "status": "expired", "ageDays": 91, "softDays": 30, "hardDays": 60, "label": "Expired for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [ { "severity": "warning", "code": "expired-source", "message": "Expired for this evidence category" } ] }, { "id": "google-parallel-search", "title": "Parallel web search grounding update", "publisher": "Google", "url": "https://blog.google/technology/ai/", "type": "Primary", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 93, "grade": "A", "authority": { "score": 40, "label": "Primary or official evidence" }, "directness": { "score": 20, "label": "Direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 8, "status": "stale", "ageDays": 91, "softDays": 90, "hardDays": 180, "label": "Review freshness before publication" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [ { "severity": "warning", "code": "stale-source", "message": "Review freshness before publication" } ] }, { "id": "deepseek-v4", "title": "DeepSeek V4 Preview", "publisher": "DeepSeek", "url": "https://api-docs.deepseek.com/news/news260424/", "type": "Primary", "category": "china-models", "critical": true, "usageCount": 3, "usageCollections": [ "benchmarkDimensions.long-context", "openModels" ], "score": 85, "grade": "B", "authority": { "score": 40, "label": "Primary or official evidence" }, "directness": { "score": 20, "label": "Direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 0, "status": "expired", "ageDays": 91, "softDays": 30, "hardDays": 60, "label": "Expired for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [ { "severity": "blocker", "code": "expired-source", "message": "Expired for this evidence category" } ] }, { "id": "deepseek-models", "title": "DeepSeek API model and alias documentation", "publisher": "DeepSeek", "url": "https://api-docs.deepseek.com/quick_start/pricing/", "type": "Primary", "category": "china-pricing", "critical": true, "usageCount": 0, "usageCollections": [], "score": 85, "grade": "B", "authority": { "score": 40, "label": "Primary or official evidence" }, "directness": { "score": 20, "label": "Direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 0, "status": "expired", "ageDays": 91, "softDays": 30, "hardDays": 60, "label": "Expired for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [ { "severity": "blocker", "code": "expired-source", "message": "Expired for this evidence category" } ] }, { "id": "amd-microsoft-helios", "title": "Microsoft to deploy AMD Helios Rackscale Solution on Azure", "publisher": "AMD", "url": "https://www.amd.com/en/newsroom/press-releases.html", "type": "Primary", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 93, "grade": "A", "authority": { "score": 40, "label": "Primary or official evidence" }, "directness": { "score": 20, "label": "Direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 8, "status": "stale", "ageDays": 91, "softDays": 90, "hardDays": 180, "label": "Review freshness before publication" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [ { "severity": "warning", "code": "stale-source", "message": "Review freshness before publication" } ] }, { "id": "reuters-waic-governance", "title": "China positions itself in global AI governance at WAIC", "publisher": "Reuters", "url": "https://www.reuters.com/technology/artificial-intelligence/", "type": "Secondary", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 71, "grade": "C", "authority": { "score": 28, "label": "Attributed secondary evidence" }, "directness": { "score": 10, "label": "Secondary reporting" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 8, "status": "stale", "ageDays": 91, "softDays": 90, "hardDays": 180, "label": "Review freshness before publication" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [ { "severity": "warning", "code": "stale-source", "message": "Review freshness before publication" }, { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review." } ] }, { "id": "reuters-huawei-waic", "title": "Huawei presents Atlas 950 SuperPoD at WAIC", "publisher": "Reuters", "url": "https://www.reuters.com/technology/artificial-intelligence/", "type": "Secondary", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 71, "grade": "C", "authority": { "score": 28, "label": "Attributed secondary evidence" }, "directness": { "score": 10, "label": "Secondary reporting" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 8, "status": "stale", "ageDays": 91, "softDays": 90, "hardDays": 180, "label": "Review freshness before publication" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [ { "severity": "warning", "code": "stale-source", "message": "Review freshness before publication" }, { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review." } ] }, { "id": "reuters-earnings-watch", "title": "Alphabet and Intel earnings put AI trade to the test", "publisher": "Reuters", "url": "https://www.reuters.com/technology/artificial-intelligence/", "type": "Secondary", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 71, "grade": "C", "authority": { "score": 28, "label": "Attributed secondary evidence" }, "directness": { "score": 10, "label": "Secondary reporting" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 8, "status": "stale", "ageDays": 91, "softDays": 90, "hardDays": 180, "label": "Review freshness before publication" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [ { "severity": "warning", "code": "stale-source", "message": "Review freshness before publication" }, { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review." } ] }, { "id": "nvidia-rtx50", "title": "GeForce RTX 50 Series announcement", "publisher": "NVIDIA", "url": "https://www.nvidia.com/en-us/geforce/graphics-cards/50-series/", "type": "Primary", "category": "hardware", "critical": false, "usageCount": 2, "usageCollections": [ "hardware" ], "score": 85, "grade": "B", "authority": { "score": 40, "label": "Primary or official evidence" }, "directness": { "score": 20, "label": "Direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 0, "status": "expired", "ageDays": 91, "softDays": 45, "hardDays": 90, "label": "Expired for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [ { "severity": "warning", "code": "expired-source", "message": "Expired for this evidence category" } ] }, { "id": "nvidia-rtx5070", "title": "GeForce RTX 5070", "publisher": "NVIDIA", "url": "https://www.nvidia.com/en-us/geforce/graphics-cards/50-series/rtx-5070-family/", "type": "Primary", "category": "hardware", "critical": true, "usageCount": 1, "usageCollections": [ "hardware" ], "score": 85, "grade": "B", "authority": { "score": 40, "label": "Primary or official evidence" }, "directness": { "score": 20, "label": "Direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 0, "status": "expired", "ageDays": 91, "softDays": 45, "hardDays": 90, "label": "Expired for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [ { "severity": "blocker", "code": "expired-source", "message": "Expired for this evidence category" } ] }, { "id": "nvidia-rtx5050", "title": "GeForce RTX 5050 announcement", "publisher": "NVIDIA", "url": "https://www.nvidia.com/en-us/geforce/news/", "type": "Primary", "category": "hardware", "critical": false, "usageCount": 1, "usageCollections": [ "hardware" ], "score": 85, "grade": "B", "authority": { "score": 40, "label": "Primary or official evidence" }, "directness": { "score": 20, "label": "Direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 0, "status": "expired", "ageDays": 91, "softDays": 45, "hardDays": 90, "label": "Expired for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [ { "severity": "warning", "code": "expired-source", "message": "Expired for this evidence category" } ] }, { "id": "amd-halo", "title": "Ryzen AI Halo Developer Platform", "publisher": "AMD / Micro Center", "url": "https://www.amd.com/en/developer/resources/ryzen-ai-software.html", "type": "Primary / retailer", "category": "hardware", "critical": false, "usageCount": 1, "usageCollections": [ "hardware" ], "score": 85, "grade": "B", "authority": { "score": 40, "label": "Primary or official evidence" }, "directness": { "score": 20, "label": "Direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 0, "status": "expired", "ageDays": 91, "softDays": 45, "hardDays": 90, "label": "Expired for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [ { "severity": "warning", "code": "expired-source", "message": "Expired for this evidence category" } ] }, { "id": "apple-m5pro", "title": "MacBook Pro with M5 Pro and M5 Max", "publisher": "Apple", "url": "https://www.apple.com/macbook-pro/", "type": "Primary", "category": "hardware", "critical": false, "usageCount": 1, "usageCollections": [ "hardware" ], "score": 85, "grade": "B", "authority": { "score": 40, "label": "Primary or official evidence" }, "directness": { "score": 20, "label": "Direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 0, "status": "expired", "ageDays": 91, "softDays": 45, "hardDays": 90, "label": "Expired for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [ { "severity": "warning", "code": "expired-source", "message": "Expired for this evidence category" } ] }, { "id": "apple-neo", "title": "MacBook Neo", "publisher": "Apple", "url": "https://www.apple.com/macbook-neo/", "type": "Primary", "category": "hardware", "critical": false, "usageCount": 1, "usageCollections": [ "hardware" ], "score": 85, "grade": "B", "authority": { "score": 40, "label": "Primary or official evidence" }, "directness": { "score": 20, "label": "Direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 0, "status": "expired", "ageDays": 91, "softDays": 45, "hardDays": 90, "label": "Expired for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [ { "severity": "warning", "code": "expired-source", "message": "Expired for this evidence category" } ] }, { "id": "market-feed", "title": "Latest available US market quote feed", "publisher": "Market data feed", "url": "https://www.nasdaq.com/market-activity", "type": "Market data", "category": "markets", "critical": false, "usageCount": 0, "usageCollections": [], "score": 74, "grade": "C", "authority": { "score": 33, "label": "Specialist data evidence" }, "directness": { "score": 16, "label": "Specialist direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 0, "status": "expired", "ageDays": 91, "softDays": 2, "hardDays": 3, "label": "Expired for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [ { "severity": "warning", "code": "expired-source", "message": "Expired for this evidence category" }, { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review." } ] }, { "id": "alibaba-qwen37", "title": "Alibaba Cloud Model Studio — Qwen flagship models", "publisher": "Alibaba Cloud", "url": "https://help.aliyun.com/en/model-studio/what-is-model-studio", "type": "Primary", "category": "china-models", "critical": true, "usageCount": 5, "usageCollections": [ "benchmarkDimensions.coding", "benchmarkDimensions.image-generation", "benchmarkDimensions.image-understanding", "benchmarkDimensions.video-audio", "models" ], "score": 85, "grade": "B", "authority": { "score": 40, "label": "Primary or official evidence" }, "directness": { "score": 20, "label": "Direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 0, "status": "expired", "ageDays": 91, "softDays": 30, "hardDays": 60, "label": "Expired for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [ { "severity": "blocker", "code": "expired-source", "message": "Expired for this evidence category" } ] }, { "id": "moonshot-kimi-k3", "title": "Kimi API platform — Kimi K3 model and pricing", "publisher": "Moonshot AI", "url": "https://platform.moonshot.cn/", "type": "Primary", "category": "china-models", "critical": true, "usageCount": 5, "usageCollections": [ "benchmarkDimensions.agents-tools", "benchmarkDimensions.coding", "benchmarkDimensions.long-context", "benchmarkDimensions.math-science", "models" ], "score": 85, "grade": "B", "authority": { "score": 40, "label": "Primary or official evidence" }, "directness": { "score": 20, "label": "Direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 0, "status": "expired", "ageDays": 91, "softDays": 30, "hardDays": 60, "label": "Expired for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [ { "severity": "blocker", "code": "expired-source", "message": "Expired for this evidence category" } ] }, { "id": "zhipu-glm52", "title": "Zhipu BigModel — GLM-5.2 model overview", "publisher": "Zhipu AI", "url": "https://docs.bigmodel.cn/cn/guide/start/model-overview", "type": "Primary", "category": "china-models", "critical": true, "usageCount": 5, "usageCollections": [ "benchmarkDimensions.agents-tools", "benchmarkDimensions.coding", "benchmarkDimensions.image-generation", "benchmarkDimensions.long-context", "models" ], "score": 85, "grade": "B", "authority": { "score": 40, "label": "Primary or official evidence" }, "directness": { "score": 20, "label": "Direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 0, "status": "expired", "ageDays": 91, "softDays": 30, "hardDays": 60, "label": "Expired for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [ { "severity": "blocker", "code": "expired-source", "message": "Expired for this evidence category" } ] }, { "id": "zhipu-glm-ocr", "title": "Zhipu BigModel — GLM-OCR", "publisher": "Zhipu AI", "url": "https://docs.bigmodel.cn/cn/guide/models/vlm/glm-ocr", "type": "Primary", "category": "benchmarks", "critical": true, "usageCount": 2, "usageCollections": [ "benchmarkDimensions.image-understanding", "benchmarkDimensions.image-understanding.specialistSignals" ], "score": 100, "grade": "A", "authority": { "score": 40, "label": "Primary or official evidence" }, "directness": { "score": 20, "label": "Direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 91, "softDays": 180, "hardDays": 365, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [] }, { "id": "baidu-ernie50", "title": "Baidu Qianfan — ERNIE 5.0 model list and international price", "publisher": "Baidu AI Cloud", "url": "https://intl.cloud.baidu.com/en/doc/qianfan/s/7m95lyy43-intl-en", "type": "Primary", "category": "china-models", "critical": true, "usageCount": 4, "usageCollections": [ "benchmarkDimensions.image-generation", "benchmarkDimensions.image-understanding", "benchmarkDimensions.video-audio", "models" ], "score": 85, "grade": "B", "authority": { "score": 40, "label": "Primary or official evidence" }, "directness": { "score": 20, "label": "Direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 0, "status": "expired", "ageDays": 91, "softDays": 30, "hardDays": 60, "label": "Expired for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [ { "severity": "blocker", "code": "expired-source", "message": "Expired for this evidence category" } ] }, { "id": "bytedance-doubao-seed21", "title": "Volcengine Ark — Doubao Seed 2.1 model catalog", "publisher": "ByteDance / Volcengine", "url": "https://www.volcengine.com/docs/82379/1330310", "type": "Primary", "category": "china-models", "critical": true, "usageCount": 1, "usageCollections": [ "models" ], "score": 85, "grade": "B", "authority": { "score": 40, "label": "Primary or official evidence" }, "directness": { "score": 20, "label": "Direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 0, "status": "expired", "ageDays": 91, "softDays": 30, "hardDays": 60, "label": "Expired for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [ { "severity": "blocker", "code": "expired-source", "message": "Expired for this evidence category" } ] }, { "id": "minimax-m3", "title": "MiniMax API — MiniMax-M3 model release and catalog", "publisher": "MiniMax", "url": "https://platform.minimax.io/docs/release-notes/models", "type": "Primary", "category": "china-models", "critical": true, "usageCount": 1, "usageCollections": [ "models" ], "score": 85, "grade": "B", "authority": { "score": 40, "label": "Primary or official evidence" }, "directness": { "score": 20, "label": "Direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 0, "status": "expired", "ageDays": 91, "softDays": 30, "hardDays": 60, "label": "Expired for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [ { "severity": "blocker", "code": "expired-source", "message": "Expired for this evidence category" } ] }, { "id": "stepfun-37-flash", "title": "StepFun — Step 3.7 Flash pricing and model documentation", "publisher": "StepFun", "url": "https://platform.stepfun.com/docs/zh/guides/pricing/details", "type": "Primary", "category": "china-models", "critical": true, "usageCount": 1, "usageCollections": [ "models" ], "score": 85, "grade": "B", "authority": { "score": 40, "label": "Primary or official evidence" }, "directness": { "score": 20, "label": "Direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 0, "status": "expired", "ageDays": 91, "softDays": 30, "hardDays": 60, "label": "Expired for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [ { "severity": "blocker", "code": "expired-source", "message": "Expired for this evidence category" } ] }, { "id": "tencent-hunyuan-a13b", "title": "Tencent Cloud — Hunyuan A13B model overview", "publisher": "Tencent Cloud", "url": "https://cloud.tencent.com/document/product/1729/104753", "type": "Primary", "category": "china-models", "critical": true, "usageCount": 1, "usageCollections": [ "models" ], "score": 85, "grade": "B", "authority": { "score": 40, "label": "Primary or official evidence" }, "directness": { "score": 20, "label": "Direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 0, "status": "expired", "ageDays": 91, "softDays": 30, "hardDays": 60, "label": "Expired for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [ { "severity": "blocker", "code": "expired-source", "message": "Expired for this evidence category" } ] }, { "id": "openai-hf-incident", "title": "OpenAI and Hugging Face partner to address security incident during model evaluation", "publisher": "OpenAI", "url": "https://openai.com/index/hugging-face-model-evaluation-security-incident/", "type": "Primary", "category": "security", "critical": true, "usageCount": 0, "usageCollections": [], "score": 85, "grade": "B", "authority": { "score": 40, "label": "Primary or official evidence" }, "directness": { "score": 20, "label": "Direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 0, "status": "expired", "ageDays": 90, "softDays": 14, "hardDays": 30, "label": "Expired for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [ { "severity": "blocker", "code": "expired-source", "message": "Expired for this evidence category" } ] }, { "id": "reuters-china-controls", "title": "China considers tighter export controls on AI models and chips, FT reports", "publisher": "Reuters", "url": "https://www.reuters.com/world/asia-pacific/china-considers-tighter-export-controls-ai-models-chips-ft-reports-2026-07-21/", "type": "Secondary reporting", "category": "china-policy", "critical": true, "usageCount": 0, "usageCollections": [], "score": 79, "grade": "C", "authority": { "score": 28, "label": "Attributed secondary evidence" }, "directness": { "score": 10, "label": "Secondary reporting" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 90, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [ { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review." } ] }, { "id": "reuters-tsmc-prices", "title": "TSMC to raise chipmaking prices by up to 10% in 2027, Nikkei Asia reports", "publisher": "Reuters", "url": "https://www.reuters.com/world/asia-pacific/tsmc-raise-chipmaking-prices-by-up-10-2027-nikkei-asia-reports-2026-07-21/", "type": "Secondary reporting", "category": "semiconductors", "critical": false, "usageCount": 0, "usageCollections": [], "score": 79, "grade": "C", "authority": { "score": 28, "label": "Attributed secondary evidence" }, "directness": { "score": 10, "label": "Secondary reporting" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 90, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [ { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review." } ] }, { "id": "supermicro-q4-update", "title": "Supermicro Provides Fourth Quarter of Fiscal Year 2026 Preliminary Business Update", "publisher": "Supermicro Investor Relations", "url": "https://ir.supermicro.com/news/news-details/2026/Supermicro-Provides-Fourth-Quarter-of-Fiscal-Year-2026-Preliminary-Business-Update/default.aspx", "type": "Primary", "category": "infrastructure", "critical": true, "usageCount": 0, "usageCollections": [], "score": 100, "grade": "A", "authority": { "score": 40, "label": "Primary or official evidence" }, "directness": { "score": 20, "label": "Direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 90, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [] }, { "id": "reuters-alphabet-earnings", "title": "Alphabet's Gemini delay, spending worries loom over earnings", "publisher": "Reuters", "url": "https://www.reuters.com/business/alphabets-gemini-delay-spending-worries-loom-over-earnings-2026-07-21/", "type": "Secondary reporting", "category": "earnings", "critical": false, "usageCount": 0, "usageCollections": [], "score": 79, "grade": "C", "authority": { "score": 28, "label": "Attributed secondary evidence" }, "directness": { "score": 10, "label": "Secondary reporting" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 90, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [ { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review." } ] }, { "id": "reuters-anthropic-patent", "title": "Anthropic sued for infringing neural network technology patents", "publisher": "Reuters", "url": "https://www.reuters.com/legal/government/university-tennessee-sues-anthropic-over-neural-network-technology-2026-07-21/", "type": "Secondary reporting", "category": "legal", "critical": false, "usageCount": 0, "usageCollections": [], "score": 71, "grade": "C", "authority": { "score": 28, "label": "Attributed secondary evidence" }, "directness": { "score": 10, "label": "Secondary reporting" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 8, "status": "stale", "ageDays": 90, "softDays": 45, "hardDays": 120, "label": "Review freshness before publication" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [ { "severity": "warning", "code": "stale-source", "message": "Review freshness before publication" }, { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review." } ] }, { "id": "reuters-humanoid-funding", "title": "Robotics startup Humanoid raises $152 million Series A round at $1.35 billion valuation", "publisher": "Reuters", "url": "https://www.reuters.com/business/robotics-startup-humanoid-raises-152-million-series-round-135-billion-valuation-2026-07-21/", "type": "Secondary reporting", "category": "physical-ai", "critical": false, "usageCount": 0, "usageCollections": [], "score": 79, "grade": "C", "authority": { "score": 28, "label": "Attributed secondary evidence" }, "directness": { "score": 10, "label": "Secondary reporting" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 90, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [ { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review." } ] }, { "id": "market-msft", "title": "Microsoft stock quote", "publisher": "Nasdaq", "url": "https://www.nasdaq.com/market-activity/stocks/msft", "type": "Market data", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 90, "grade": "A", "authority": { "score": 33, "label": "Specialist data evidence" }, "directness": { "score": 16, "label": "Specialist direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 90, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [] }, { "id": "market-googl", "title": "Alphabet Class A stock quote", "publisher": "Nasdaq", "url": "https://www.nasdaq.com/market-activity/stocks/googl", "type": "Market data", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 90, "grade": "A", "authority": { "score": 33, "label": "Specialist data evidence" }, "directness": { "score": 16, "label": "Specialist direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 90, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [] }, { "id": "market-meta", "title": "Meta Platforms stock quote", "publisher": "Nasdaq", "url": "https://www.nasdaq.com/market-activity/stocks/meta", "type": "Market data", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 90, "grade": "A", "authority": { "score": 33, "label": "Specialist data evidence" }, "directness": { "score": 16, "label": "Specialist direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 90, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [] }, { "id": "market-amzn", "title": "Amazon stock quote", "publisher": "Nasdaq", "url": "https://www.nasdaq.com/market-activity/stocks/amzn", "type": "Market data", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 90, "grade": "A", "authority": { "score": 33, "label": "Specialist data evidence" }, "directness": { "score": 16, "label": "Specialist direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 90, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [] }, { "id": "market-tsm", "title": "TSMC ADR stock quote", "publisher": "NYSE", "url": "https://www.nyse.com/quote/XNYS:TSM", "type": "Market data", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 90, "grade": "A", "authority": { "score": 33, "label": "Specialist data evidence" }, "directness": { "score": 16, "label": "Specialist direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 90, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [] }, { "id": "market-orcl", "title": "Oracle stock quote", "publisher": "NYSE", "url": "https://www.nyse.com/quote/XNYS:ORCL", "type": "Market data", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 90, "grade": "A", "authority": { "score": 33, "label": "Specialist data evidence" }, "directness": { "score": 16, "label": "Specialist direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 90, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [] }, { "id": "openai-presence", "title": "Introducing OpenAI Presence", "publisher": "OpenAI", "url": "https://openai.com/index/introducing-openai-presence/", "type": "Primary", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 100, "grade": "A", "authority": { "score": 40, "label": "Primary or official evidence" }, "directness": { "score": 20, "label": "Direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 88, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [] }, { "id": "openai-national-science", "title": "Advancing the next era of national science", "publisher": "OpenAI", "url": "https://openai.com/index/advancing-the-next-era-of-national-science/", "type": "Primary", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 100, "grade": "A", "authority": { "score": 40, "label": "Primary or official evidence" }, "directness": { "score": 20, "label": "Direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 88, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [] }, { "id": "amd-anthropic", "title": "AMD and Anthropic announce strategic partnership for up to 2 GW of MI450 Series GPUs", "publisher": "AMD Newsroom", "url": "https://newsroom.amd.com/news/amd-anthropic-strategic-partnership/", "type": "Primary", "category": "benchmarks", "critical": true, "usageCount": 1, "usageCollections": [ "benchmarkDimensions.efficiency-cost" ], "score": 100, "grade": "A", "authority": { "score": 40, "label": "Primary or official evidence" }, "directness": { "score": 20, "label": "Direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 88, "softDays": 180, "hardDays": 365, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [] }, { "id": "google-q2-2026", "title": "Alphabet Q2 2026 earnings call — CEO remarks", "publisher": "Google / Alphabet", "url": "https://blog.google/company-news/inside-google/message-ceo/alphabet-earnings-q2-2026/", "type": "Primary", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 100, "grade": "A", "authority": { "score": 40, "label": "Primary or official evidence" }, "directness": { "score": 20, "label": "Direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 88, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [] }, { "id": "qwen-image3", "title": "Qwen-Image-3.0: Rich Content, Authentic Details, Deep Knowledge", "publisher": "Alibaba Cloud / Qwen", "url": "https://www.alibabacloud.com/blog/qwen-image-3-0-rich-content-authentic-details-deep-knowledge_603385", "type": "Primary provider publication", "category": "models", "critical": true, "usageCount": 1, "usageCollections": [ "models" ], "score": 85, "grade": "B", "authority": { "score": 40, "label": "Primary or official evidence" }, "directness": { "score": 20, "label": "Direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 0, "status": "expired", "ageDays": 88, "softDays": 30, "hardDays": 60, "label": "Expired for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [ { "severity": "blocker", "code": "expired-source", "message": "Expired for this evidence category" } ] }, { "id": "qwen-audio3", "title": "Qwen-Audio-3.0-TTS: More Multilingual, Easier to Direct", "publisher": "Alibaba Cloud / Qwen", "url": "https://www.alibabacloud.com/blog/qwen-audio-3-0-tts-more-multilingual-easier-to-direct_603379", "type": "Primary provider publication", "category": "benchmarks", "critical": true, "usageCount": 2, "usageCollections": [ "benchmarkDimensions.efficiency-cost", "models" ], "score": 100, "grade": "A", "authority": { "score": 40, "label": "Primary or official evidence" }, "directness": { "score": 20, "label": "Direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 88, "softDays": 180, "hardDays": 365, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [] }, { "id": "reuters-server-cpu-china", "title": "Intel and AMD sign longer-term server CPU deals with Chinese clients as prices surge", "publisher": "Reuters", "url": "https://www.reuters.com/legal/transactional/intel-amd-sign-long-term-server-cpu-deals-with-chinese-clients-prices-surge-2026-07-23/", "type": "Secondary reporting", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 79, "grade": "C", "authority": { "score": 28, "label": "Attributed secondary evidence" }, "directness": { "score": 10, "label": "Secondary reporting" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 88, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [ { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review." } ] }, { "id": "ap-markets-jul22", "title": "How major US stock indexes fared Wednesday, July 22, 2026", "publisher": "Associated Press", "url": "https://apnews.com/article/87e0cd1ef1994f0fcbfb8f025064f64f", "type": "Market summary", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 79, "grade": "C", "authority": { "score": 28, "label": "Attributed secondary evidence" }, "directness": { "score": 10, "label": "Secondary reporting" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 88, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [ { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review." } ] }, { "id": "market-nvda", "title": "NVIDIA July 22, 2026 closing quote", "publisher": "Yahoo Finance", "url": "https://finance.yahoo.com/quote/NVDA/", "type": "Market data", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 90, "grade": "A", "authority": { "score": 33, "label": "Specialist data evidence" }, "directness": { "score": 16, "label": "Specialist direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 88, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [] }, { "id": "market-amd", "title": "AMD July 22, 2026 closing quote", "publisher": "ChartExchange", "url": "https://chartexchange.com/symbol/nasdaq-amd/", "type": "Market data", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 90, "grade": "A", "authority": { "score": 33, "label": "Specialist data evidence" }, "directness": { "score": 16, "label": "Specialist direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 88, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [] }, { "id": "market-smci", "title": "Super Micro Computer July 22, 2026 closing quote", "publisher": "ChartExchange", "url": "https://chartexchange.com/symbol/nasdaq-smci/historical/", "type": "Market data", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 90, "grade": "A", "authority": { "score": 33, "label": "Specialist data evidence" }, "directness": { "score": 16, "label": "Specialist direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 88, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [] }, { "id": "market-avgo", "title": "Broadcom July 22, 2026 closing quote", "publisher": "MarketWatch", "url": "https://www.marketwatch.com/investing/stock/avgo", "type": "Market data", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 90, "grade": "A", "authority": { "score": 33, "label": "Specialist data evidence" }, "directness": { "score": 16, "label": "Specialist direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 88, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [] }, { "id": "microsoft-copilot-pro", "title": "Microsoft Copilot Pro plan and pricing", "publisher": "Microsoft", "url": "https://www.microsoft.com/en-us/store/b/copilotpro", "type": "Primary plan page", "category": "pricing", "critical": false, "usageCount": 1, "usageCollections": [ "plans" ], "score": 85, "grade": "B", "authority": { "score": 40, "label": "Primary or official evidence" }, "directness": { "score": 20, "label": "Direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 0, "status": "expired", "ageDays": 88, "softDays": 30, "hardDays": 60, "label": "Expired for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [ { "severity": "warning", "code": "expired-source", "message": "Expired for this evidence category" } ] }, { "id": "microsoft-copilot-business-promo", "title": "Microsoft 365 Copilot plans and promotional pricing", "publisher": "Microsoft", "url": "https://www.microsoft.com/en-us/microsoft-365-copilot/pricing", "type": "Primary promotion page", "category": "pricing", "critical": false, "usageCount": 1, "usageCollections": [ "promotions" ], "score": 85, "grade": "B", "authority": { "score": 40, "label": "Primary or official evidence" }, "directness": { "score": 20, "label": "Direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 0, "status": "expired", "ageDays": 88, "softDays": 30, "hardDays": 60, "label": "Expired for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [ { "severity": "warning", "code": "expired-source", "message": "Expired for this evidence category" } ] }, { "id": "perplexity-plans", "title": "Which Perplexity subscription plan is right for you?", "publisher": "Perplexity Support", "url": "https://www.perplexity.ai/help-center/en/articles/11187416-which-perplexity-subscription-plan-is-right-for-you", "type": "Primary plan documentation", "category": "pricing", "critical": false, "usageCount": 2, "usageCollections": [ "freeTiers", "plans" ], "score": 85, "grade": "B", "authority": { "score": 40, "label": "Primary or official evidence" }, "directness": { "score": 20, "label": "Direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 0, "status": "expired", "ageDays": 88, "softDays": 30, "hardDays": 60, "label": "Expired for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [ { "severity": "warning", "code": "expired-source", "message": "Expired for this evidence category" } ] }, { "id": "xai-grok-plans", "title": "Grok product and plan page", "publisher": "xAI", "url": "https://grok.com/", "type": "Primary product page", "category": "pricing", "critical": false, "usageCount": 1, "usageCollections": [ "plans" ], "score": 85, "grade": "B", "authority": { "score": 40, "label": "Primary or official evidence" }, "directness": { "score": 20, "label": "Direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 0, "status": "expired", "ageDays": 88, "softDays": 30, "hardDays": 60, "label": "Expired for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [ { "severity": "warning", "code": "expired-source", "message": "Expired for this evidence category" } ] }, { "id": "anthropic-sonnet5", "title": "Introducing Claude Sonnet 5", "publisher": "Anthropic", "url": "https://www.anthropic.com/news/claude-sonnet-5", "type": "Primary launch and pricing page", "category": "pricing", "critical": false, "usageCount": 1, "usageCollections": [ "promotions" ], "score": 85, "grade": "B", "authority": { "score": 40, "label": "Primary or official evidence" }, "directness": { "score": 20, "label": "Direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 0, "status": "expired", "ageDays": 88, "softDays": 30, "hardDays": 60, "label": "Expired for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [ { "severity": "warning", "code": "expired-source", "message": "Expired for this evidence category" } ] }, { "id": "openai-health-chatgpt", "title": "Launching Health in ChatGPT", "publisher": "OpenAI", "url": "https://openai.com/index/health-in-chatgpt/", "type": "Primary", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 100, "grade": "A", "authority": { "score": 40, "label": "Primary or official evidence" }, "directness": { "score": 20, "label": "Direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 88, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [] }, { "id": "google-atlas-ai-economy", "title": "Understanding the AI economy — the first ATLAS report", "publisher": "Google", "url": "https://blog.google/innovation-and-ai/technology/research/understanding-the-ai-economy/", "type": "Primary", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 100, "grade": "A", "authority": { "score": 40, "label": "Primary or official evidence" }, "directness": { "score": 20, "label": "Direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 88, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [] }, { "id": "deepseek-v4-pricing-retirement", "title": "DeepSeek V4 models, pricing and legacy alias retirement", "publisher": "DeepSeek API Docs", "url": "https://api-docs.deepseek.com/quick_start/pricing", "type": "Primary", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 100, "grade": "A", "authority": { "score": 40, "label": "Primary or official evidence" }, "directness": { "score": 20, "label": "Direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 88, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [] }, { "id": "ftc-ai-accuracy-policy", "title": "FTC seeks public comment on policy statement addressing AI accuracy", "publisher": "Federal Trade Commission", "url": "https://www.ftc.gov/news-events/news/press-releases/2026/07/ftc-seeks-public-comment-policy-statement-addressing-ai-accuracy", "type": "Government source", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 100, "grade": "A", "authority": { "score": 40, "label": "Primary or official evidence" }, "directness": { "score": 20, "label": "Direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 88, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [] }, { "id": "ap-markets-jul23", "title": "How major US stock indexes fared Thursday, July 23, 2026", "publisher": "Associated Press", "url": "https://apnews.com/article/wall-street-dow-nasdaq-stocks-3cb4bc150707e0a4378312fbb825f5e4", "type": "Market summary", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 79, "grade": "C", "authority": { "score": 28, "label": "Attributed secondary evidence" }, "directness": { "score": 10, "label": "Secondary reporting" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 88, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [ { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review." } ] }, { "id": "market-googl-jul23", "title": "Alphabet July 23, 2026 closing quote", "publisher": "Nasdaq", "url": "https://www.nasdaq.com/market-activity/stocks/googl/historical", "type": "Historical market data", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 90, "grade": "A", "authority": { "score": 33, "label": "Specialist data evidence" }, "directness": { "score": 16, "label": "Specialist direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 88, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [] }, { "id": "market-nvda-jul23", "title": "NVIDIA July 23, 2026 closing quote", "publisher": "Nasdaq", "url": "https://www.nasdaq.com/market-activity/stocks/nvda/historical", "type": "Historical market data", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 90, "grade": "A", "authority": { "score": 33, "label": "Specialist data evidence" }, "directness": { "score": 16, "label": "Specialist direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 88, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [] }, { "id": "market-amd-jul23", "title": "AMD July 23, 2026 closing quote", "publisher": "Nasdaq", "url": "https://www.nasdaq.com/market-activity/stocks/amd/historical", "type": "Historical market data", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 90, "grade": "A", "authority": { "score": 33, "label": "Specialist data evidence" }, "directness": { "score": 16, "label": "Specialist direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 88, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [] }, { "id": "market-smci-jul23", "title": "Super Micro Computer July 23, 2026 closing quote", "publisher": "Nasdaq", "url": "https://www.nasdaq.com/market-activity/stocks/smci/historical", "type": "Historical market data", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 90, "grade": "A", "authority": { "score": 33, "label": "Specialist data evidence" }, "directness": { "score": 16, "label": "Specialist direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 88, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [] }, { "id": "market-avgo-jul23", "title": "Broadcom July 23, 2026 closing quote", "publisher": "Nasdaq", "url": "https://www.nasdaq.com/market-activity/stocks/avgo/historical", "type": "Historical market data", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 90, "grade": "A", "authority": { "score": 33, "label": "Specialist data evidence" }, "directness": { "score": 16, "label": "Specialist direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 88, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [] }, { "id": "market-msft-jul23", "title": "Microsoft July 23, 2026 closing quote", "publisher": "Nasdaq", "url": "https://www.nasdaq.com/market-activity/stocks/msft/historical", "type": "Historical market data", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 90, "grade": "A", "authority": { "score": 33, "label": "Specialist data evidence" }, "directness": { "score": 16, "label": "Specialist direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 88, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [] }, { "id": "meta-ai-actions-jul24", "title": "Meta AI Doesn’t Just Think, It Acts", "publisher": "Meta", "url": "https://about.fb.com/news/2026/07/meta-ai-muse-spark-doesnt-just-think-it-acts/", "type": "Primary", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 100, "grade": "A", "authority": { "score": 40, "label": "Primary or official evidence" }, "directness": { "score": 20, "label": "Direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 86, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [] }, { "id": "google-eu-ai-transparency-code", "title": "Google signs the EU AI Act Code of Practice on Transparency of AI-Generated Content", "publisher": "Google", "url": "https://blog.google/company-news/outreach-and-initiatives/public-policy/eu-ai-act-transparency-code-of-practice/", "type": "Primary", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 100, "grade": "A", "authority": { "score": 40, "label": "Primary or official evidence" }, "directness": { "score": 20, "label": "Direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 86, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [] }, { "id": "amd-advancing-ai-2026", "title": "AMD Advancing AI 2026", "publisher": "AMD", "url": "https://www.amd.com/en/corporate/events/advancing-ai.html", "type": "Primary", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 100, "grade": "A", "authority": { "score": 40, "label": "Primary or official evidence" }, "directness": { "score": 20, "label": "Direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 86, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [] }, { "id": "deepseek-v4-post-retirement", "title": "DeepSeek V4 models, pricing and legacy alias retirement", "publisher": "DeepSeek API Docs", "url": "https://api-docs.deepseek.com/quick_start/pricing", "type": "Primary", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 100, "grade": "A", "authority": { "score": 40, "label": "Primary or official evidence" }, "directness": { "score": 20, "label": "Direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 86, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [] }, { "id": "market-summary-jul24", "title": "Broadcom shares fall as the S&P 500 and Dow finish slightly higher", "publisher": "MarketWatch", "url": "https://www.marketwatch.com/data-news/broadcom-inc-stock-underperforms-friday-when-compared-to-competitors-9e8602a4-a84725bbedab", "type": "Market summary", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 79, "grade": "C", "authority": { "score": 28, "label": "Attributed secondary evidence" }, "directness": { "score": 10, "label": "Secondary reporting" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 86, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [ { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review." } ] }, { "id": "market-googl-jul24", "title": "Alphabet July 24, 2026 historical quote", "publisher": "Nasdaq", "url": "https://www.nasdaq.com/market-activity/stocks/googl/historical", "type": "Historical market data", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 90, "grade": "A", "authority": { "score": 33, "label": "Specialist data evidence" }, "directness": { "score": 16, "label": "Specialist direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 86, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [] }, { "id": "market-nvda-jul24", "title": "NVIDIA July 24, 2026 historical quote", "publisher": "Nasdaq", "url": "https://www.nasdaq.com/market-activity/stocks/nvda/historical", "type": "Historical market data", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 90, "grade": "A", "authority": { "score": 33, "label": "Specialist data evidence" }, "directness": { "score": 16, "label": "Specialist direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 86, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [] }, { "id": "market-amd-jul24", "title": "AMD July 24, 2026 historical quote", "publisher": "Nasdaq", "url": "https://www.nasdaq.com/market-activity/stocks/amd/historical", "type": "Historical market data", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 90, "grade": "A", "authority": { "score": 33, "label": "Specialist data evidence" }, "directness": { "score": 16, "label": "Specialist direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 86, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [] }, { "id": "market-smci-jul24", "title": "Super Micro Computer July 24, 2026 historical quote", "publisher": "Nasdaq", "url": "https://www.nasdaq.com/market-activity/stocks/smci/historical", "type": "Historical market data", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 90, "grade": "A", "authority": { "score": 33, "label": "Specialist data evidence" }, "directness": { "score": 16, "label": "Specialist direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 86, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [] }, { "id": "market-avgo-jul24", "title": "Broadcom July 24, 2026 historical quote", "publisher": "Nasdaq", "url": "https://www.nasdaq.com/market-activity/stocks/avgo/historical", "type": "Historical market data", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 90, "grade": "A", "authority": { "score": 33, "label": "Specialist data evidence" }, "directness": { "score": 16, "label": "Specialist direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 86, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [] }, { "id": "nvidia-sk-ai-factory-jul24", "title": "SK Group and NVIDIA Expand Strategic Partnership Across AI Factories and Next-Generation Memory", "publisher": "NVIDIA Newsroom", "url": "https://nvidianews.nvidia.com/news/sk-group-and-nvidia-expand-strategic-partnership-across-ai-factories-and-next-generation-memory", "type": "Primary", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 100, "grade": "A", "authority": { "score": 40, "label": "Primary or official evidence" }, "directness": { "score": 20, "label": "Direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 85, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [] }, { "id": "nvidia-korea-summit-jul24", "title": "At AI Summit, South Korea Outlines Its AI Future With NVIDIA and Partners", "publisher": "NVIDIA Blog", "url": "https://blogs.nvidia.com/blog/ai-summit-korea-partners-and-nvidia/", "type": "Primary", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 100, "grade": "A", "authority": { "score": 40, "label": "Primary or official evidence" }, "directness": { "score": 20, "label": "Direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 85, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [] }, { "id": "nvidia-kaist-lab-jul23", "title": "NVIDIA and KAIST Launch Joint AI Research Lab to Accelerate AI Innovation in Korea", "publisher": "NVIDIA Newsroom", "url": "https://nvidianews.nvidia.com/news/nvidia-and-kaist-launch-joint-ai-research-lab-to-accelerate-ai-innovation-in-korea", "type": "Primary", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 100, "grade": "A", "authority": { "score": 40, "label": "Primary or official evidence" }, "directness": { "score": 20, "label": "Direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 85, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [] }, { "id": "wsj-openai-ohio-backstop-jul26", "title": "Nvidia in Talks With OpenAI to Guarantee $250 Billion Financing for Data Center", "publisher": "The Wall Street Journal", "url": "https://www.wsj.com/tech/ai/nvidia-in-talks-with-openai-to-guarantee-250-billion-financing-for-data-center-3dd6eae3", "type": "Secondary reporting", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 79, "grade": "C", "authority": { "score": 28, "label": "Attributed secondary evidence" }, "directness": { "score": 10, "label": "Secondary reporting" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 85, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [ { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review." } ] }, { "id": "reuters-openai-ohio-backstop-jul26", "title": "Nvidia in talks with OpenAI to guarantee $250 billion financing for data center, WSJ reports", "publisher": "Reuters", "url": "https://wtaq.com/2026/07/26/nvidia-in-talks-with-openai-to-guarantee-250-billion-financing-for-data-center-wsj-reports/", "type": "Secondary reporting", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 79, "grade": "C", "authority": { "score": 28, "label": "Attributed secondary evidence" }, "directness": { "score": 10, "label": "Secondary reporting" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 85, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [ { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review." } ] }, { "id": "theinformation-openai-ohio-campus-jun09", "title": "OpenAI in Talks to Lease 10 Gigawatt Ohio Data Center with Backing From Nvidia", "publisher": "The Information", "url": "https://www.theinformation.com/articles/openai-talks-lease-10-gigawatt-ohio-data-center-backing-nvidia/", "type": "Secondary reporting", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 79, "grade": "C", "authority": { "score": 28, "label": "Attributed secondary evidence" }, "directness": { "score": 10, "label": "Secondary reporting" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 85, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [ { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review." } ] }, { "id": "nvidia-open-secure-ai-alliance-jul27", "title": "Industry Leaders Unite in Open Secure AI Alliance for AI Safety and Security", "publisher": "NVIDIA Blog", "url": "https://blogs.nvidia.com/blog/open-secure-ai-alliance/", "type": "Primary", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 100, "grade": "A", "authority": { "score": 40, "label": "Primary or official evidence" }, "directness": { "score": 20, "label": "Direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 84, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [] }, { "id": "nvidia-ssi-partnership-jul27", "title": "Ilya Sutskever’s Safe Superintelligence Inc. and NVIDIA Announce Long-Term Strategic Partnership", "publisher": "NVIDIA Newsroom", "url": "https://nvidianews.nvidia.com/news/ilya-sutskevers-safe-superintelligence-inc-and-nvidia-announce-long-term-strategic-partnership", "type": "Primary", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 100, "grade": "A", "authority": { "score": 40, "label": "Primary or official evidence" }, "directness": { "score": 20, "label": "Direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 84, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [] }, { "id": "openai-work-frontier-jul27", "title": "How AI is expanding what people do at work", "publisher": "OpenAI", "url": "https://openai.com/index/how-ai-is-expanding-what-people-do-at-work/", "type": "Primary", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 100, "grade": "A", "authority": { "score": 40, "label": "Primary or official evidence" }, "directness": { "score": 20, "label": "Direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 84, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [] }, { "id": "reuters-altman-huang-warner-jul27", "title": "OpenAI CEO Sam Altman and NVIDIA CEO Jensen Huang to meet Senator Mark Warner", "publisher": "Reuters", "url": "https://www.reuters.com/business/openais-sam-altman-meet-with-senate-intelligence-committees-top-democrat-2026-07-27/", "type": "Secondary reporting", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 79, "grade": "C", "authority": { "score": 28, "label": "Attributed secondary evidence" }, "directness": { "score": 10, "label": "Secondary reporting" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 84, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [ { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review." } ] }, { "id": "ap-us-market-close-jul27", "title": "How major US stock indexes fared Monday 7/27/2026", "publisher": "The Associated Press", "url": "https://apnews.com/article/wall-street-stocks-dow-nasdaq-e7325919557efc52535a96f09ab38b0d", "type": "Market summary", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 79, "grade": "C", "authority": { "score": 28, "label": "Attributed secondary evidence" }, "directness": { "score": 10, "label": "Secondary reporting" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 84, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [ { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review." } ] }, { "id": "market-nvda-jul27", "title": "NVIDIA July 27, 2026 historical quote", "publisher": "Investing.com", "url": "https://www.investing.com/equities/nvidia-corp-historical-data?symbol=NVDA", "type": "Historical market data", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 90, "grade": "A", "authority": { "score": 33, "label": "Specialist data evidence" }, "directness": { "score": 16, "label": "Specialist direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 84, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [] }, { "id": "meta-blackrock-el-paso-jul28", "title": "Meta Announces New Strategic Venture with BlackRock to Develop Data Center in El Paso", "publisher": "Meta Investor Relations", "url": "https://investor.atmeta.com/investor-news/press-release-details/2026/Meta-Announces-New-Strategic-Venture-with-BlackRock-to-Develop-Data-Center-in-El-Paso/default.aspx", "type": "Primary", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 100, "grade": "A", "authority": { "score": 40, "label": "Primary or official evidence" }, "directness": { "score": 20, "label": "Direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 82, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [] }, { "id": "pacing-frontier-jul28", "title": "Pacing the Frontier", "publisher": "Pacing the Frontier", "url": "https://www.pacingthefrontier.com/", "type": "Employee statement", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 71, "grade": "C", "authority": { "score": 22, "label": "Unclassified attributed evidence" }, "directness": { "score": 8, "label": "Indirect or unclassified evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 82, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [ { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review." } ] }, { "id": "mcp-releases-status-jul28", "title": "Releases · modelcontextprotocol/modelcontextprotocol", "publisher": "Model Context Protocol on GitHub", "url": "https://github.com/modelcontextprotocol/modelcontextprotocol/releases", "type": "Official repository", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 100, "grade": "A", "authority": { "score": 40, "label": "Primary or official evidence" }, "directness": { "score": 20, "label": "Direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 82, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [] }, { "id": "mcp-versioning-status-jul28", "title": "Versioning - Model Context Protocol", "publisher": "Model Context Protocol", "url": "https://modelcontextprotocol.io/docs/learn/versioning", "type": "Official documentation", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 100, "grade": "A", "authority": { "score": 40, "label": "Primary or official evidence" }, "directness": { "score": 20, "label": "Direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 82, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [] }, { "id": "fcc-robots-inverters-jul28", "title": "FCC bans humanoid robots from China, citing unacceptable risks", "publisher": "The Guardian", "url": "https://www.theguardian.com/us-news/2026/jul/28/fcc-ban-humanoid-robots-china", "type": "Secondary reporting with FCC determination link", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 79, "grade": "C", "authority": { "score": 28, "label": "Attributed secondary evidence" }, "directness": { "score": 10, "label": "Secondary reporting" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 82, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [ { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review." } ] }, { "id": "reuters-chip-rout-jul28", "title": "Chip rout", "publisher": "Reuters", "url": "https://www.reuters.com/podcasts/reuters-morning-bid/chip-rout-2026-07-28/", "type": "Market analysis", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 71, "grade": "C", "authority": { "score": 22, "label": "Unclassified attributed evidence" }, "directness": { "score": 8, "label": "Indirect or unclassified evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 82, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [ { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review." } ] }, { "id": "ap-us-market-close-jul28", "title": "How major US stock indexes fared Tuesday 7/28/2026", "publisher": "The Associated Press", "url": "https://apnews.com/article/wall-street-stocks-dow-nasdaq-d2a114bce818cab6ee9583f4815cb89a", "type": "Market summary", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 79, "grade": "C", "authority": { "score": 28, "label": "Attributed secondary evidence" }, "directness": { "score": 10, "label": "Secondary reporting" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 82, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [ { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review." } ] }, { "id": "market-qcom-jul28", "title": "Qualcomm Inc. stock underperforms Tuesday when compared to competitors", "publisher": "MarketWatch", "url": "https://www.marketwatch.com/data-news/qualcomm-inc-stock-underperforms-tuesday-when-compared-to-competitors-55f7faa8-4edc212b9770", "type": "Historical market data", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 90, "grade": "A", "authority": { "score": 33, "label": "Specialist data evidence" }, "directness": { "score": 16, "label": "Specialist direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 82, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [] }, { "id": "reuters-wall-street-fed-ai-jul29", "title": "Wall Street closes down sharply after Fed holds rates unchanged", "publisher": "Reuters", "url": "https://www.reuters.com/business/sp-500-nasdaq-futures-inch-up-before-fed-decision-chip-stocks-wobble-2026-07-29/", "type": "Market and earnings reporting", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 71, "grade": "C", "authority": { "score": 22, "label": "Unclassified attributed evidence" }, "directness": { "score": 8, "label": "Indirect or unclassified evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 82, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [ { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review." } ] }, { "id": "ap-us-market-close-jul29", "title": "How major US stock indexes fared Wednesday 7/29/2026", "publisher": "The Associated Press", "url": "https://apnews.com/article/a483060a0aa9c24c50f589b0b7713497", "type": "Market summary", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 79, "grade": "C", "authority": { "score": 28, "label": "Attributed secondary evidence" }, "directness": { "score": 10, "label": "Secondary reporting" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 82, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [ { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review." } ] }, { "id": "meta-q2-event-jul29", "title": "Q2 2026 Earnings Call", "publisher": "Meta Investor Relations", "url": "https://investor.atmeta.com/investor-events/event-details/2026/Q2-2026-Earnings-Call/default.aspx", "type": "Primary", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 100, "grade": "A", "authority": { "score": 40, "label": "Primary or official evidence" }, "directness": { "score": 20, "label": "Direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 82, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [] }, { "id": "microsoft-fy26-q4-event-jul29", "title": "Microsoft announces quarterly earnings release date", "publisher": "Microsoft Source", "url": "https://news.microsoft.com/source/2026/07/08/microsoft-announces-quarterly-earnings-release-date-68/", "type": "Primary", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 100, "grade": "A", "authority": { "score": 40, "label": "Primary or official evidence" }, "directness": { "score": 20, "label": "Direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 82, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [] }, { "id": "mcp-releases-jul29", "title": "Releases · modelcontextprotocol/modelcontextprotocol", "publisher": "Model Context Protocol on GitHub", "url": "https://github.com/modelcontextprotocol/modelcontextprotocol/releases", "type": "Official repository", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 100, "grade": "A", "authority": { "score": 40, "label": "Primary or official evidence" }, "directness": { "score": 20, "label": "Direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 82, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [] }, { "id": "mcp-versioning-jul29", "title": "Versioning - Model Context Protocol", "publisher": "Model Context Protocol", "url": "https://modelcontextprotocol.io/docs/learn/versioning", "type": "Official documentation", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 100, "grade": "A", "authority": { "score": 40, "label": "Primary or official evidence" }, "directness": { "score": 20, "label": "Direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 82, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [] }, { "id": "eu-ai-transparency-guidelines-jul20", "title": "Commission publishes guidelines on transparency obligations for providers and deployers of certain AI systems", "publisher": "European Commission", "url": "https://digital-strategy.ec.europa.eu/en/news/commission-publishes-guidelines-transparency-obligations-providers-and-deployers-certain-ai-systems", "type": "Primary", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 100, "grade": "A", "authority": { "score": 40, "label": "Primary or official evidence" }, "directness": { "score": 20, "label": "Direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 82, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [] }, { "id": "microsoft-project-perception-jul27", "title": "Rethinking security for the age of AI", "publisher": "Microsoft", "url": "https://blogs.microsoft.com/blog/2026/07/27/rethinking-security-for-the-age-of-ai/", "type": "Primary", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 100, "grade": "A", "authority": { "score": 40, "label": "Primary or official evidence" }, "directness": { "score": 20, "label": "Direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 82, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [] }, { "id": "market-spx-jul29", "title": "S&P 500 Index historical data", "publisher": "MarketWatch", "url": "https://www.marketwatch.com/investing/index/spx/historical", "type": "Historical market data", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 90, "grade": "A", "authority": { "score": 33, "label": "Specialist data evidence" }, "directness": { "score": 16, "label": "Specialist direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 82, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [] }, { "id": "market-dji-jul29", "title": "Dow Jones Industrial Average historical data", "publisher": "Yahoo Finance", "url": "https://finance.yahoo.com/quote/%5EDJI/history/", "type": "Historical market data", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 90, "grade": "A", "authority": { "score": 33, "label": "Specialist data evidence" }, "directness": { "score": 16, "label": "Specialist direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 82, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [] }, { "id": "market-ixic-jul29", "title": "Nasdaq Composite historical data", "publisher": "Yahoo Finance", "url": "https://finance.yahoo.com/quote/%5EIXIC/history/", "type": "Historical market data", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 90, "grade": "A", "authority": { "score": 33, "label": "Specialist data evidence" }, "directness": { "score": 16, "label": "Specialist direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 82, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [] }, { "id": "reuters-amazon-aws-jul30", "title": "Amazon lifts investment plans after strong cloud sales; shares jump", "publisher": "Reuters", "url": "https://www.reuters.com/business/retail-consumer/amazon-beats-estimates-quarterly-cloud-revenue-growth-2026-07-30/", "type": "Market and earnings reporting", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 71, "grade": "C", "authority": { "score": 22, "label": "Unclassified attributed evidence" }, "directness": { "score": 8, "label": "Indirect or unclassified evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 81, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [ { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review." } ] }, { "id": "reuters-wall-street-microsoft-jul30", "title": "Wall Street ends sharply higher, lifted by soaring Microsoft", "publisher": "Reuters", "url": "https://www.reuters.com/business/us-stock-futures-steady-microsoft-offsets-fed-concerns-2026-07-30/", "type": "Market and earnings reporting", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 71, "grade": "C", "authority": { "score": 22, "label": "Unclassified attributed evidence" }, "directness": { "score": 8, "label": "Indirect or unclassified evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 81, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [ { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review." } ] }, { "id": "ap-apple-q3-jul30", "title": "Strong iPhone sales propel Apple fiscal Q3 results past Wall Street expectations", "publisher": "The Associated Press", "url": "https://apnews.com/article/94102918cb3592ebc1d2a38c4d7d819a", "type": "Company earnings reporting", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 71, "grade": "C", "authority": { "score": 22, "label": "Unclassified attributed evidence" }, "directness": { "score": 8, "label": "Indirect or unclassified evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 81, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [ { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review." } ] }, { "id": "amazon-q2-event-jul30", "title": "Q2 2026 Amazon.com earnings conference call", "publisher": "Amazon Investor Relations", "url": "https://ir.aboutamazon.com/events/event-details/default.aspx", "type": "Official company event", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 100, "grade": "A", "authority": { "score": 40, "label": "Primary or official evidence" }, "directness": { "score": 20, "label": "Direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 81, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [] }, { "id": "mcp-releases-jul30", "title": "Releases · modelcontextprotocol/modelcontextprotocol", "publisher": "Model Context Protocol on GitHub", "url": "https://github.com/modelcontextprotocol/modelcontextprotocol/releases", "type": "Official repository", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 100, "grade": "A", "authority": { "score": 40, "label": "Primary or official evidence" }, "directness": { "score": 20, "label": "Direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 81, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [] }, { "id": "mcp-versioning-jul30", "title": "Versioning - Model Context Protocol 2026-07-28", "publisher": "Model Context Protocol", "url": "https://modelcontextprotocol.io/docs/2026-07-28/learn/versioning", "type": "Official documentation", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 100, "grade": "A", "authority": { "score": 40, "label": "Primary or official evidence" }, "directness": { "score": 20, "label": "Direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 81, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [] }, { "id": "market-spx-jul30", "title": "S&P 500 Index historical data", "publisher": "MarketWatch", "url": "https://www.marketwatch.com/investing/index/spx/historical", "type": "Historical market data", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 90, "grade": "A", "authority": { "score": 33, "label": "Specialist data evidence" }, "directness": { "score": 16, "label": "Specialist direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 81, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [] }, { "id": "market-dji-jul30", "title": "Dow Jones Industrial Average historical data", "publisher": "Yahoo Finance", "url": "https://finance.yahoo.com/quote/%5EDJI/history/", "type": "Historical market data", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 90, "grade": "A", "authority": { "score": 33, "label": "Specialist data evidence" }, "directness": { "score": 16, "label": "Specialist direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 81, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [] }, { "id": "market-ixic-jul30", "title": "Nasdaq Composite historical data", "publisher": "Yahoo Finance", "url": "https://finance.yahoo.com/quote/%5EIXIC/history/", "type": "Historical market data", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 90, "grade": "A", "authority": { "score": 33, "label": "Specialist data evidence" }, "directness": { "score": 16, "label": "Specialist direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 81, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [] }, { "id": "eu-article50-aug2", "title": "Commission publishes guidelines on transparency obligations for providers and deployers of certain AI systems", "publisher": "European Commission", "url": "https://digital-strategy.ec.europa.eu/en/news/commission-publishes-guidelines-transparency-obligations-providers-and-deployers-certain-ai-systems", "type": "Official regulatory guidance", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 100, "grade": "A", "authority": { "score": 40, "label": "Primary or official evidence" }, "directness": { "score": 20, "label": "Direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 78, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [] }, { "id": "eu-gpai-enforcement-aug2", "title": "Commission enforcement powers for obligations covering the most advanced general-purpose AI models", "publisher": "European Commission AI Act Service Desk", "url": "https://ai-act-service-desk.ec.europa.eu/en/faq?faq_category_id=69", "type": "Official enforcement guidance", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 100, "grade": "A", "authority": { "score": 40, "label": "Primary or official evidence" }, "directness": { "score": 20, "label": "Direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 78, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [] }, { "id": "eu-ai-act-timeline-aug2", "title": "Timeline for the implementation of the EU AI Act", "publisher": "European Commission AI Act Service Desk", "url": "https://ai-act-service-desk.ec.europa.eu/en/ai-act/timeline/timeline-implementation-eu-ai-act", "type": "Official implementation timeline", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 100, "grade": "A", "authority": { "score": 40, "label": "Primary or official evidence" }, "directness": { "score": 20, "label": "Direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 78, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [] }, { "id": "eu-governance-enforcement-aug2", "title": "Governance and enforcement of the AI Act", "publisher": "European Commission", "url": "https://digital-strategy.ec.europa.eu/en/policies/ai-act-governance-and-enforcement", "type": "Official governance guidance", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 100, "grade": "A", "authority": { "score": 40, "label": "Primary or official evidence" }, "directness": { "score": 20, "label": "Direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 78, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [] }, { "id": "huggingface-agent-incident-jul16", "title": "Security incident disclosure — July 2026", "publisher": "Hugging Face", "url": "https://huggingface.co/blog/security-incident-july-2026", "type": "Official security disclosure", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 100, "grade": "A", "authority": { "score": 40, "label": "Primary or official evidence" }, "directness": { "score": 20, "label": "Direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 78, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [] }, { "id": "reuters-eu-agent-talks-jul31", "title": "EU in talks with OpenAI, Anthropic after rogue AI agent hacks", "publisher": "Reuters", "url": "https://www.reuters.com/world/eu-says-necessary-monitor-high-risk-ai-systems-after-openai-anthropic-ai-hacking-2026-07-31/", "type": "Regulatory and security reporting", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 71, "grade": "C", "authority": { "score": 22, "label": "Unclassified attributed evidence" }, "directness": { "score": 8, "label": "Indirect or unclassified evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 78, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [ { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review." } ] }, { "id": "reuters-us-voluntary-tests-aug3", "title": "Meta, Anthropic, Google, OpenAI to meet Trump officials about AI safety testing", "publisher": "Reuters", "url": "https://www.reuters.com/world/us-finalizes-voluntary-ai-safety-tests-white-house-official-says-2026-08-03/", "type": "Government and technology reporting", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 100, "grade": "A", "authority": { "score": 40, "label": "Primary or official evidence" }, "directness": { "score": 20, "label": "Direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 76, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [] }, { "id": "whitehouse-frontier-security-order-jun2", "title": "Promoting Advanced Artificial Intelligence Innovation and Security", "publisher": "The White House", "url": "https://www.whitehouse.gov/presidential-actions/2026/06/promoting-advanced-artificial-intelligence-innovation-and-security/", "type": "Official executive order", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 100, "grade": "A", "authority": { "score": 40, "label": "Primary or official evidence" }, "directness": { "score": 20, "label": "Direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 76, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [] }, { "id": "reuters-uk-voluntary-regulation-aug3", "title": "Britain says it is open to AI regulation if voluntary safeguards fall short", "publisher": "Reuters", "url": "https://www.reuters.com/business/media-telecom/britain-says-it-is-open-ai-regulation-if-voluntary-safeguards-fall-short-2026-08-03/", "type": "Government and technology reporting", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 100, "grade": "A", "authority": { "score": 40, "label": "Primary or official evidence" }, "directness": { "score": 20, "label": "Direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 76, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [] }, { "id": "uk-ai-cyber-code-jul10", "title": "Code of Practice for the Cyber Security of AI", "publisher": "UK Government", "url": "https://www.gov.uk/government/publications/ai-cyber-security-code-of-practice/code-of-practice-for-the-cyber-security-of-ai", "type": "Official cyber-security guidance", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 100, "grade": "A", "authority": { "score": 40, "label": "Primary or official evidence" }, "directness": { "score": 20, "label": "Direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 76, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [] }, { "id": "reuters-palantir-q2-aug3", "title": "Palantir lifts annual revenue forecast on steady demand for AI-powered data analytics", "publisher": "Reuters", "url": "https://www.reuters.com/technology/palantir-raises-annual-revenue-forecast-strong-demand-us-government-commercial-2026-08-03/", "type": "Company earnings reporting", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 71, "grade": "C", "authority": { "score": 22, "label": "Unclassified attributed evidence" }, "directness": { "score": 8, "label": "Indirect or unclassified evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 76, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [ { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review." } ] }, { "id": "ap-us-markets-aug3", "title": "How major US stock indexes fared Monday 8/3/2026", "publisher": "The Associated Press", "url": "https://apnews.com/article/wall-street-stocks-dow-nasdaq-f8e5f81b45c83878f5b7f69832bb0c95", "type": "Market close reporting", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 71, "grade": "C", "authority": { "score": 22, "label": "Unclassified attributed evidence" }, "directness": { "score": 8, "label": "Indirect or unclassified evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 76, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [ { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review." } ] }, { "id": "reuters-ai-futures-aug4", "title": "US stock futures advance as strong AI-linked earnings continue", "publisher": "Reuters", "url": "https://www.reuters.com/business/nasdaq-futures-underpinned-by-strong-ai-forecasts-focus-earnings-data-2026-08-04/", "type": "Premarket and earnings reporting", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 71, "grade": "C", "authority": { "score": 22, "label": "Unclassified attributed evidence" }, "directness": { "score": 8, "label": "Indirect or unclassified evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 76, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [ { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review." } ] }, { "id": "market-spx-aug3", "title": "S&P 500 Index historical data", "publisher": "MarketWatch", "url": "https://www.marketwatch.com/investing/index/spx/historical", "type": "Historical market data", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 90, "grade": "A", "authority": { "score": 33, "label": "Specialist data evidence" }, "directness": { "score": 16, "label": "Specialist direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 76, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [] }, { "id": "market-dji-aug3", "title": "Dow Jones Industrial Average historical data", "publisher": "Yahoo Finance", "url": "https://finance.yahoo.com/quote/%5EDJI/history/", "type": "Historical market data", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 90, "grade": "A", "authority": { "score": 33, "label": "Specialist data evidence" }, "directness": { "score": 16, "label": "Specialist direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 76, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [] }, { "id": "market-ixic-aug3", "title": "Nasdaq Composite historical data", "publisher": "Yahoo Finance", "url": "https://finance.yahoo.com/quote/%5EIXIC/history/", "type": "Historical market data", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 90, "grade": "A", "authority": { "score": 33, "label": "Specialist data evidence" }, "directness": { "score": 16, "label": "Specialist direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 76, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [] }, { "id": "reuters-whitehouse-open-weight-aug4", "title": "Trump advisers tell AI firms they will not safety-test open-weight models", "publisher": "Reuters", "url": "https://www.reuters.com/legal/litigation/meta-anthropic-google-openai-meet-with-trump-white-house-amid-rogue-ai-agent-2026-08-04/", "type": "Government and technology reporting", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 100, "grade": "A", "authority": { "score": 40, "label": "Primary or official evidence" }, "directness": { "score": 20, "label": "Direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 76, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [] }, { "id": "aisi-unsanctioned-agent-report-aug4", "title": "Incident Report: unsanctioned agent behaviour during cyber testing", "publisher": "UK AI Security Institute", "url": "https://www.aisi.gov.uk/blog/incident-report-unsanctioned-agent-behaviour-during-cyber-testing", "type": "Official government incident report", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 100, "grade": "A", "authority": { "score": 40, "label": "Primary or official evidence" }, "directness": { "score": 20, "label": "Direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 76, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [] }, { "id": "openai-third-party-cyber-evals-aug4", "title": "Third-party cyber evaluations involving OpenAI models", "publisher": "OpenAI", "url": "https://openai.com/index/third-party-cyber-evaluations-involving-openai-models/", "type": "Official company security disclosure", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 100, "grade": "A", "authority": { "score": 40, "label": "Primary or official evidence" }, "directness": { "score": 20, "label": "Direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 76, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [] }, { "id": "reuters-agent-breaches-aug5", "title": "OpenAI, Anthropic AI agents implicated in new security breaches", "publisher": "Reuters", "url": "https://www.reuters.com/legal/litigation/openai-anthropic-ai-agents-implicated-new-security-breaches-2026-08-05/", "type": "Cybersecurity and technology reporting", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 71, "grade": "C", "authority": { "score": 22, "label": "Unclassified attributed evidence" }, "directness": { "score": 8, "label": "Indirect or unclassified evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 76, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [ { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review." } ] }, { "id": "reuters-anthropic-global-affairs-aug4", "title": "Anthropic names global affairs chief to tackle AI policy as Trump tensions persist", "publisher": "Reuters", "url": "https://www.reuters.com/world/anthropic-names-global-affairs-chief-tackle-ai-policy-trump-tensions-persist-2026-08-04/", "type": "Company and policy reporting", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 71, "grade": "C", "authority": { "score": 22, "label": "Unclassified attributed evidence" }, "directness": { "score": 8, "label": "Indirect or unclassified evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 76, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [ { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review." } ] }, { "id": "reuters-us-markets-aug4", "title": "Dow, S&P 500 close at record on AI-linked earnings, Mideast deal hopes", "publisher": "Reuters", "url": "https://www.reuters.com/business/nasdaq-futures-underpinned-by-strong-ai-forecasts-focus-earnings-data-2026-08-04/", "type": "Market close and company earnings reporting", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 71, "grade": "C", "authority": { "score": 22, "label": "Unclassified attributed evidence" }, "directness": { "score": 8, "label": "Indirect or unclassified evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 76, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [ { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review." } ] }, { "id": "market-spx-aug4", "title": "S&P 500 Index historical data — August 4, 2026 close", "publisher": "MarketWatch", "url": "https://www.marketwatch.com/investing/index/spx/historical", "type": "Historical market data", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 90, "grade": "A", "authority": { "score": 33, "label": "Specialist data evidence" }, "directness": { "score": 16, "label": "Specialist direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 76, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [] }, { "id": "market-dji-aug4", "title": "Dow Jones Industrial Average historical data — August 4, 2026 close", "publisher": "Yahoo Finance", "url": "https://finance.yahoo.com/quote/%5EDJI/history/", "type": "Historical market data", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 90, "grade": "A", "authority": { "score": 33, "label": "Specialist data evidence" }, "directness": { "score": 16, "label": "Specialist direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 76, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [] }, { "id": "market-ixic-aug4", "title": "Nasdaq Composite historical data — August 4, 2026 close", "publisher": "Yahoo Finance", "url": "https://finance.yahoo.com/quote/%5EIXIC/history/", "type": "Historical market data", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 90, "grade": "A", "authority": { "score": 33, "label": "Specialist data evidence" }, "directness": { "score": 16, "label": "Specialist direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 76, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [] }, { "id": "reuters-google-ai-leadership-aug5", "title": "Google shakes up AI leadership as DeepMind chief shifts role", "publisher": "Reuters", "url": "https://www.reuters.com/business/google-shakes-up-ai-leadership-deepmind-chief-shifts-role-2026-08-05/", "type": "Company and technology reporting", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 71, "grade": "C", "authority": { "score": 22, "label": "Unclassified attributed evidence" }, "directness": { "score": 8, "label": "Indirect or unclassified evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 75, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [ { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review." } ] }, { "id": "reuters-meta-cyber-eval-aug5", "title": "Meta AI model hacks another company during testing", "publisher": "Reuters", "url": "https://www.reuters.com/technology/metas-ai-model-hacked-another-company-during-testing-information-reports-2026-08-05/", "type": "Cybersecurity and technology reporting", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 71, "grade": "C", "authority": { "score": 22, "label": "Unclassified attributed evidence" }, "directness": { "score": 8, "label": "Indirect or unclassified evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 75, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [ { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review." } ] }, { "id": "reuters-amd-ai-payoff-aug5", "title": "AMD falls as investors demand bigger AI payoff", "publisher": "Reuters", "url": "https://www.reuters.com/business/amd-falls-investors-seek-bigger-ai-payoff-2026-08-05/", "type": "Semiconductor and market reporting", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 71, "grade": "C", "authority": { "score": 22, "label": "Unclassified attributed evidence" }, "directness": { "score": 8, "label": "Indirect or unclassified evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 75, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [ { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review." } ] }, { "id": "reuters-global-markets-aug5", "title": "Stocks mixed with Nasdaq down; US oil eases on Iran talks", "publisher": "Reuters", "url": "https://www.reuters.com/world/china/global-markets-global-markets-2026-08-05/", "type": "Market close reporting", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 71, "grade": "C", "authority": { "score": 22, "label": "Unclassified attributed evidence" }, "directness": { "score": 8, "label": "Indirect or unclassified evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 75, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [ { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review." } ] }, { "id": "market-spx-aug5", "title": "S&P 500 Index historical data — August 5, 2026 close", "publisher": "MarketWatch", "url": "https://www.marketwatch.com/investing/index/spx/historical", "type": "Historical market data", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 90, "grade": "A", "authority": { "score": 33, "label": "Specialist data evidence" }, "directness": { "score": 16, "label": "Specialist direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 75, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [] }, { "id": "market-dji-aug5", "title": "Dow Jones Industrial Average historical data — August 5, 2026 close", "publisher": "Yahoo Finance", "url": "https://finance.yahoo.com/quote/%5EDJI/history/", "type": "Historical market data", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 90, "grade": "A", "authority": { "score": 33, "label": "Specialist data evidence" }, "directness": { "score": 16, "label": "Specialist direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 75, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [] }, { "id": "market-ixic-aug5", "title": "Nasdaq Composite historical data — August 5, 2026 close", "publisher": "Yahoo Finance", "url": "https://finance.yahoo.com/quote/%5EIXIC/history/", "type": "Historical market data", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 90, "grade": "A", "authority": { "score": 33, "label": "Specialist data evidence" }, "directness": { "score": 16, "label": "Specialist direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 75, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [] }, { "id": "reuters-alibaba-qwen38-license-aug7", "title": "Alibaba plans to charge big users of its next open-source AI model, sources say", "publisher": "Reuters", "url": "https://www.reuters.com/business/retail-consumer/alibaba-plans-charge-big-users-its-next-open-source-ai-model-sources-say-2026-08-07/", "type": "Company, licensing and AI-market reporting", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 71, "grade": "C", "authority": { "score": 22, "label": "Unclassified attributed evidence" }, "directness": { "score": 8, "label": "Indirect or unclassified evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 73, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [ { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review." } ] }, { "id": "reuters-amd-taalas-aug6", "title": "AMD deepens AI inference bet with Taalas deal as chip race heats up", "publisher": "Reuters", "url": "https://www.reuters.com/business/amd-deepens-ai-inference-bet-with-taalas-deal-chip-race-heats-up-2026-08-06/", "type": "Semiconductor and acquisition reporting", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 71, "grade": "C", "authority": { "score": 22, "label": "Unclassified attributed evidence" }, "directness": { "score": 8, "label": "Indirect or unclassified evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 73, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [ { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review." } ] }, { "id": "reuters-deepseek-unitree-aug6", "title": "DeepSeek invests $20.8 million in Unitree's Shanghai IPO", "publisher": "Reuters", "url": "https://www.reuters.com/world/asia-pacific/deepseek-invests-208-million-unitrees-shanghai-ipo-2026-08-06/", "type": "AI, robotics and China technology reporting", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 71, "grade": "C", "authority": { "score": 22, "label": "Unclassified attributed evidence" }, "directness": { "score": 8, "label": "Indirect or unclassified evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 73, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [ { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review." } ] }, { "id": "reuters-deepseek-funding-aug6", "title": "DeepSeek resumes funding round seeking nearly $8 billion, Bloomberg News reports", "publisher": "Reuters", "url": "https://www.reuters.com/world/asia-pacific/deepseek-resumes-funding-round-seeking-nearly-8-billion-bloomberg-news-reports-2026-08-06/", "type": "Company funding and valuation reporting", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 71, "grade": "C", "authority": { "score": 22, "label": "Unclassified attributed evidence" }, "directness": { "score": 8, "label": "Indirect or unclassified evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 73, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [ { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review." } ] }, { "id": "reuters-us-ai-policy-pressure-aug6", "title": "Trump's tech ties come under bipartisan fire after AI agents go rogue", "publisher": "Reuters", "url": "https://www.reuters.com/legal/litigation/trumps-tech-ties-blamed-ai-inaction-by-maga-republicans-democrats-2026-08-06/", "type": "AI policy and cybersecurity reporting", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 71, "grade": "C", "authority": { "score": 22, "label": "Unclassified attributed evidence" }, "directness": { "score": 8, "label": "Indirect or unclassified evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 73, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [ { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review." } ] }, { "id": "reuters-us-markets-aug6", "title": "Stocks close lower as investors eye Mideast talks, earnings", "publisher": "Reuters", "url": "https://www.reuters.com/business/retail-consumer/sp-500-dow-futures-steady-mideast-deal-focus-chips-stumble-2026-08-06/", "type": "Market close and company earnings reporting", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 71, "grade": "C", "authority": { "score": 22, "label": "Unclassified attributed evidence" }, "directness": { "score": 8, "label": "Indirect or unclassified evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 73, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [ { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review." } ] }, { "id": "reuters-global-markets-aug7", "title": "Asian shares pause for US jobs, oil extends gains on Mideast risk", "publisher": "Reuters", "url": "https://www.reuters.com/world/china/global-markets-global-markets-2026-08-07/", "type": "Global market and macroeconomic reporting", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 71, "grade": "C", "authority": { "score": 22, "label": "Unclassified attributed evidence" }, "directness": { "score": 8, "label": "Indirect or unclassified evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 73, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [ { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review." } ] }, { "id": "market-spx-aug6", "title": "S&P 500 Index historical data — August 6, 2026 close", "publisher": "MarketWatch", "url": "https://www.marketwatch.com/investing/index/spx/historical", "type": "Historical market data", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 90, "grade": "A", "authority": { "score": 33, "label": "Specialist data evidence" }, "directness": { "score": 16, "label": "Specialist direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 73, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [] }, { "id": "market-dji-aug6", "title": "Dow Jones Industrial Average historical data — August 6, 2026 close", "publisher": "Yahoo Finance", "url": "https://finance.yahoo.com/quote/%5EDJI/history/", "type": "Historical market data", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 90, "grade": "A", "authority": { "score": 33, "label": "Specialist data evidence" }, "directness": { "score": 16, "label": "Specialist direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 73, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [] }, { "id": "market-ixic-aug6", "title": "Nasdaq Composite historical data — August 6, 2026 close", "publisher": "Yahoo Finance", "url": "https://finance.yahoo.com/quote/%5EIXIC/history/", "type": "Historical market data", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 90, "grade": "A", "authority": { "score": 33, "label": "Specialist data evidence" }, "directness": { "score": 16, "label": "Specialist direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 73, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [] }, { "id": "whitehouse-frontier-ai-eo-jun2", "title": "Promoting Advanced Artificial Intelligence Innovation and Security", "publisher": "The White House", "url": "https://www.whitehouse.gov/presidential-actions/2026/06/promoting-advanced-artificial-intelligence-innovation-and-security/", "type": "Primary government policy source", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 100, "grade": "A", "authority": { "score": 40, "label": "Primary or official evidence" }, "directness": { "score": 20, "label": "Direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 72, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [] }, { "id": "guardian-whitehouse-ai-framework-aug7", "title": "The White House's plan to vet potentially dangerous AI is cloaked in secrecy", "publisher": "The Guardian", "url": "https://www.theguardian.com/technology/2026/aug/07/white-house-ai", "type": "AI policy and transparency reporting", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 71, "grade": "C", "authority": { "score": 22, "label": "Unclassified attributed evidence" }, "directness": { "score": 8, "label": "Indirect or unclassified evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 72, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [ { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review." } ] }, { "id": "reuters-open-weight-testing-aug4", "title": "Trump advisers tell AI firms they will not safety-test open-weight models", "publisher": "Reuters", "url": "https://www.reuters.com/legal/litigation/meta-anthropic-google-openai-meet-with-trump-white-house-amid-rogue-ai-agent-2026-08-04/", "type": "AI policy and cybersecurity reporting", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 71, "grade": "C", "authority": { "score": 22, "label": "Unclassified attributed evidence" }, "directness": { "score": 8, "label": "Indirect or unclassified evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 72, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [ { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review." } ] }, { "id": "reuters-us-jobs-aug7", "title": "US suffers unexpected job losses in July, markets dial back rate hike expectations", "publisher": "Reuters", "url": "https://www.reuters.com/business/us-nonfarm-payrolls-fall-july-unemployment-rate-eases-41-2026-08-07/", "type": "U.S. labor-market and macroeconomic reporting", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 71, "grade": "C", "authority": { "score": 22, "label": "Unclassified attributed evidence" }, "directness": { "score": 8, "label": "Indirect or unclassified evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 72, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [ { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review." } ] }, { "id": "reuters-us-markets-aug7", "title": "S&P closes at record high as soft jobs report eases rate-hike concerns", "publisher": "Reuters", "url": "https://www.reuters.com/business/sp-500-dow-futures-muted-ahead-jobs-data-chips-software-stocks-rise-2026-08-07/", "type": "U.S. market close and rate-expectation reporting", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 71, "grade": "C", "authority": { "score": 22, "label": "Unclassified attributed evidence" }, "directness": { "score": 8, "label": "Indirect or unclassified evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 72, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [ { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review." } ] }, { "id": "reuters-inflation-weekahead-aug7", "title": "Wall St Week Ahead: Inflation data to test record-setting US stocks, Fed rate views", "publisher": "Reuters", "url": "https://www.reuters.com/business/wall-st-week-ahead-inflation-data-test-record-setting-us-stocks-fed-rate-views-2026-08-07/", "type": "Markets, inflation and technology-sector reporting", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 71, "grade": "C", "authority": { "score": 22, "label": "Unclassified attributed evidence" }, "directness": { "score": 8, "label": "Indirect or unclassified evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 72, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [ { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review." } ] }, { "id": "wsj-source-foundry-aug7", "title": "Situational Awareness Bets $400 Million on Stealth Chip Startup After Crash", "publisher": "The Wall Street Journal", "url": "https://www.wsj.com/tech/ai/situational-awareness-bets-400-million-on-stealth-chip-startup-after-crash-02c7374e", "type": "AI infrastructure and semiconductor investment reporting", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 71, "grade": "C", "authority": { "score": 22, "label": "Unclassified attributed evidence" }, "directness": { "score": 8, "label": "Indirect or unclassified evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 72, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [ { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review." } ] }, { "id": "market-dji-aug7", "title": "Dow Jones Industrial Average historical data — August 7, 2026 close", "publisher": "Yahoo Finance", "url": "https://finance.yahoo.com/quote/%5EDJI/history/", "type": "Historical market data", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 90, "grade": "A", "authority": { "score": 33, "label": "Specialist data evidence" }, "directness": { "score": 16, "label": "Specialist direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 72, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [] }, { "id": "market-ixic-aug7", "title": "Nasdaq Composite historical data — August 7, 2026 close", "publisher": "Yahoo Finance", "url": "https://finance.yahoo.com/quote/%5EIXIC/history/", "type": "Historical market data", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 90, "grade": "A", "authority": { "score": 33, "label": "Specialist data evidence" }, "directness": { "score": 16, "label": "Specialist direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 72, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [] }, { "id": "market-spx-aug7", "title": "S&P 500 Index historical data — August 7, 2026 close", "publisher": "MarketWatch", "url": "https://www.marketwatch.com/investing/index/spx/historical", "type": "Historical market data", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 90, "grade": "A", "authority": { "score": 33, "label": "Specialist data evidence" }, "directness": { "score": 16, "label": "Specialist direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 72, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [] }, { "id": "reuters-apple-qwen-mac-aug8", "title": "Apple says Mac users in China can connect to Alibaba's Qwen AI service", "publisher": "Reuters", "url": "https://www.reuters.com/business/retail-consumer/apple-says-mac-users-china-can-connect-alibabas-qwen-ai-service-2026-08-08/", "type": "AI product and China-market reporting", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 71, "grade": "C", "authority": { "score": 22, "label": "Unclassified attributed evidence" }, "directness": { "score": 8, "label": "Indirect or unclassified evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 72, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [ { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review." } ] }, { "id": "reuters-openai-astra-cyber-aug7", "title": "OpenAI flags possible critical cybersecurity risk in upcoming model, tightens controls", "publisher": "Reuters", "url": "https://www.reuters.com/legal/litigation/openai-flags-possible-critical-cybersecurity-risk-upcoming-model-tightens-2026-08-07/", "type": "Frontier-model cybersecurity reporting", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 71, "grade": "C", "authority": { "score": 22, "label": "Unclassified attributed evidence" }, "directness": { "score": 8, "label": "Indirect or unclassified evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 72, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [ { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review." } ] }, { "id": "openai-preparedness-framework-v2", "title": "Preparedness Framework v2", "publisher": "OpenAI", "url": "https://cdn.openai.com/pdf/18a02b5d-6b67-4cec-ab64-68cdfbddebcd/preparedness-framework-v2.pdf", "type": "Primary AI safety framework", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 100, "grade": "A", "authority": { "score": 40, "label": "Primary or official evidence" }, "directness": { "score": 20, "label": "Direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 72, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [] }, { "id": "reuters-aisi-agent-evals-aug5", "title": "OpenAI, Anthropic AI agents implicated in new security breaches", "publisher": "Reuters", "url": "https://www.reuters.com/legal/litigation/openai-anthropic-ai-agents-implicated-new-security-breaches-2026-08-05/", "type": "AI agent evaluation and cybersecurity reporting", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 71, "grade": "C", "authority": { "score": 22, "label": "Unclassified attributed evidence" }, "directness": { "score": 8, "label": "Indirect or unclassified evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 72, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [ { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review." } ] }, { "id": "reuters-nist-ai-evals-aug7", "title": "Trump says Congress wants to regulate AI industry 'out of business'", "publisher": "Reuters", "url": "https://www.reuters.com/legal/litigation/trump-says-congress-wants-regulate-ai-industry-out-business-2026-08-07/", "type": "U.S. AI policy and standards reporting", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 71, "grade": "C", "authority": { "score": 22, "label": "Unclassified attributed evidence" }, "directness": { "score": 8, "label": "Indirect or unclassified evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 72, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [ { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review." } ] }, { "id": "apple-cn-intelligence-support-aug8", "title": "Use Apple Intelligence on Mac", "publisher": "Apple Support China", "url": "https://support.apple.com/zh-cn/guide/mac-help-cn/mchl46361784/26/mac/26", "type": "Primary product support documentation", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 100, "grade": "A", "authority": { "score": 40, "label": "Primary or official evidence" }, "directness": { "score": 20, "label": "Direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 72, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [] }, { "id": "openai-atlas-retirement-aug9", "title": "Evolving Atlas into ChatGPT for browser-based agentic work", "publisher": "OpenAI Help Center", "url": "https://help.openai.com/en/articles/20001371-evolving-atlas-into-chatgpt-for-browser-based-agentic-work", "type": "Primary product documentation", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 100, "grade": "A", "authority": { "score": 40, "label": "Primary or official evidence" }, "directness": { "score": 20, "label": "Direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 71, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [] }, { "id": "anthropic-opus5-release-jul24", "title": "Claude Platform release notes — Claude Opus 5", "publisher": "Anthropic", "url": "https://platform.claude.com/docs/en/release-notes/overview", "type": "Primary model documentation", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 100, "grade": "A", "authority": { "score": 40, "label": "Primary or official evidence" }, "directness": { "score": 20, "label": "Direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 71, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [] }, { "id": "arcprize-opus5-jul24", "title": "Claude Opus 5 — ARC-AGI results", "publisher": "ARC Prize", "url": "https://arcprize.org/results/anthropic-claude-opus-5", "type": "Benchmark-owner result page", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 93, "grade": "A", "authority": { "score": 36, "label": "Research or benchmark evidence" }, "directness": { "score": 16, "label": "Specialist direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 71, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [] }, { "id": "arcprize-gpt56-jul9", "title": "GPT-5.6 — ARC-AGI results", "publisher": "ARC Prize", "url": "https://arcprize.org/results/openai-gpt-5-6", "type": "Benchmark-owner result page", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 93, "grade": "A", "authority": { "score": 36, "label": "Research or benchmark evidence" }, "directness": { "score": 16, "label": "Specialist direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 71, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [] }, { "id": "reuters-qwen38-aug3", "title": "Alibaba unveils Qwen3.8-Max", "publisher": "Reuters", "url": "https://www.reuters.com/business/retail-consumer/alibaba-unveils-its-most-capable-ai-model-date-not-far-behind-moonshots-size-2026-08-03/", "type": "Model release reporting", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 71, "grade": "C", "authority": { "score": 22, "label": "Unclassified attributed evidence" }, "directness": { "score": 8, "label": "Indirect or unclassified evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 71, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [ { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review." } ] }, { "id": "alibaba-model-catalog-aug9", "title": "Alibaba Cloud Model Studio model catalog", "publisher": "Alibaba Cloud", "url": "https://www.alibabacloud.com/help/en/model-studio/models", "type": "Primary model catalog", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 100, "grade": "A", "authority": { "score": 40, "label": "Primary or official evidence" }, "directness": { "score": 20, "label": "Direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 71, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [] }, { "id": "alibaba-model-pricing-aug9", "title": "Alibaba Cloud Model Studio pricing", "publisher": "Alibaba Cloud", "url": "https://www.alibabacloud.com/help/en/model-studio/model-pricing", "type": "Primary pricing documentation", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 100, "grade": "A", "authority": { "score": 40, "label": "Primary or official evidence" }, "directness": { "score": 20, "label": "Direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 71, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [] }, { "id": "deepseek-v4-apr24", "title": "DeepSeek V4-Pro and V4-Flash release notes", "publisher": "DeepSeek", "url": "https://api-docs.deepseek.com/news/news260424/", "type": "Primary model release documentation", "category": "models", "critical": true, "usageCount": 1, "usageCollections": [ "models" ], "score": 85, "grade": "B", "authority": { "score": 40, "label": "Primary or official evidence" }, "directness": { "score": 20, "label": "Direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 0, "status": "expired", "ageDays": 71, "softDays": 30, "hardDays": 60, "label": "Expired for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [ { "severity": "blocker", "code": "expired-source", "message": "Expired for this evidence category" } ] }, { "id": "deepseek-v4-pricing-aug9", "title": "DeepSeek API pricing — V4-Flash", "publisher": "DeepSeek", "url": "https://api-docs.deepseek.com/quick_start/pricing/", "type": "Primary API pricing documentation", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 100, "grade": "A", "authority": { "score": 40, "label": "Primary or official evidence" }, "directness": { "score": 20, "label": "Direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 71, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [] }, { "id": "artificialanalysis-v4flash-aug9", "title": "DeepSeek V4 Flash — independent model analysis", "publisher": "Artificial Analysis", "url": "https://artificialanalysis.ai/models/deepseek-v4-flash", "type": "Independent model evaluation", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 71, "grade": "C", "authority": { "score": 22, "label": "Unclassified attributed evidence" }, "directness": { "score": 8, "label": "Indirect or unclassified evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 71, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [ { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review." } ] }, { "id": "reuters-deepseek-v4-aug3", "title": "DeepSeek V4-Flash cost and performance reporting", "publisher": "Reuters", "url": "https://www.reuters.com/business/retail-consumer/deepseeks-new-ai-model-is-by-far-cheapest-well-known-models-run-research-firm-2026-08-03/", "type": "Model economics reporting", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 71, "grade": "C", "authority": { "score": 22, "label": "Unclassified attributed evidence" }, "directness": { "score": 8, "label": "Indirect or unclassified evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 71, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [ { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review." } ] }, { "id": "reuters-bigtech-leases-aug4", "title": "AI data-centre race builds $1 trillion lease burden for Big Tech", "publisher": "Reuters", "url": "https://www.reuters.com/business/retail-consumer/ai-data-centre-race-builds-1-trillion-lease-burden-big-tech-2026-08-04/", "type": "Infrastructure and financial reporting", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 71, "grade": "C", "authority": { "score": 22, "label": "Unclassified attributed evidence" }, "directness": { "score": 8, "label": "Indirect or unclassified evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 71, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [ { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review." } ] }, { "id": "sec-msft-leases-mar31", "title": "Microsoft Form 10-Q — leases not yet commenced", "publisher": "U.S. Securities and Exchange Commission", "url": "https://www.sec.gov/Archives/edgar/data/789019/000119312526191507/msft-20260331.htm", "type": "Primary financial filing", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 100, "grade": "A", "authority": { "score": 40, "label": "Primary or official evidence" }, "directness": { "score": 20, "label": "Direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 71, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [] }, { "id": "sec-meta-leases-mar31", "title": "Meta Platforms Form 10-Q lease commitments", "publisher": "U.S. Securities and Exchange Commission", "url": "https://www.sec.gov/Archives/edgar/data/1326801/000162828026028526/R16.htm", "type": "Primary financial filing", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 100, "grade": "A", "authority": { "score": 40, "label": "Primary or official evidence" }, "directness": { "score": 20, "label": "Direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 71, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [] }, { "id": "sec-alphabet-leases-mar31", "title": "Alphabet Form 10-Q — leases not yet commenced", "publisher": "U.S. Securities and Exchange Commission", "url": "https://www.sec.gov/Archives/edgar/data/1652044/000165204426000048/goog-20260331.htm", "type": "Primary financial filing", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 100, "grade": "A", "authority": { "score": 40, "label": "Primary or official evidence" }, "directness": { "score": 20, "label": "Direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 71, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [] }, { "id": "aisi-test-time-compute", "title": "Measuring AI agents progress on multi-step cyber attack scenarios", "publisher": "UK AI Security Institute", "url": "https://www.aisi.gov.uk/research/measuring-ai-agents-progress-on-multi-step-cyber-attack-scenarios", "type": "Government AI evaluation research", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 100, "grade": "A", "authority": { "score": 40, "label": "Primary or official evidence" }, "directness": { "score": 20, "label": "Direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 71, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [] }, { "id": "anthropic-models-overview-aug9", "title": "Claude models overview", "publisher": "Anthropic", "url": "https://platform.claude.com/docs/en/about-claude/models/overview", "type": "Primary model documentation", "category": "models", "critical": true, "usageCount": 1, "usageCollections": [ "models" ], "score": 85, "grade": "B", "authority": { "score": 40, "label": "Primary or official evidence" }, "directness": { "score": 20, "label": "Direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 0, "status": "expired", "ageDays": 71, "softDays": 30, "hardDays": 60, "label": "Expired for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [ { "severity": "blocker", "code": "expired-source", "message": "Expired for this evidence category" } ] }, { "id": "reuters-nvidia-financing-aug10", "title": "Wall Street giants partner with Nvidia on $500 billion AI financing plan", "publisher": "Reuters", "url": "https://www.reuters.com/technology/wall-street-giants-partner-with-nvidia-500-billion-ai-financing-deal-ft-reports-2026-08-10/", "type": "AI infrastructure financing reporting", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 71, "grade": "C", "authority": { "score": 22, "label": "Unclassified attributed evidence" }, "directness": { "score": 8, "label": "Indirect or unclassified evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 70, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [ { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review." } ] }, { "id": "nvidia-naver-brookfield-jul24", "title": "NAVER, NVIDIA and Brookfield to Expand Korea’s National AI Factory Infrastructure Buildout", "publisher": "NVIDIA Newsroom", "url": "https://nvidianews.nvidia.com/news/naver-nvidia-and-brookfield-to-expand-koreas-national-ai-factory-infrastructure-buildout", "type": "Primary infrastructure announcement", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 100, "grade": "A", "authority": { "score": 40, "label": "Primary or official evidence" }, "directness": { "score": 20, "label": "Direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 70, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [] }, { "id": "reuters-datacenter-lender-risk-aug10", "title": "Lenders scrutinize U.S. data-center financing as community opposition builds", "publisher": "Reuters", "url": "https://www.reuters.com/legal/transactional/lenders-scrutinize-us-data-center-financing-community-opposition-builds-2026-08-10/", "type": "Infrastructure financing and permitting reporting", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 71, "grade": "C", "authority": { "score": 22, "label": "Unclassified attributed evidence" }, "directness": { "score": 8, "label": "Indirect or unclassified evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 70, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [ { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review." } ] }, { "id": "reuters-house-agent-oversight-aug10", "title": "U.S. House Democrats press Anthropic, OpenAI about AI-agent incidents", "publisher": "Reuters", "url": "https://www.reuters.com/legal/litigation/us-house-democrats-press-anthropic-openai-about-rogue-ai-agents-2026-08-10/", "type": "U.S. policy and AI-agent safety reporting", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 71, "grade": "C", "authority": { "score": 22, "label": "Unclassified attributed evidence" }, "directness": { "score": 8, "label": "Indirect or unclassified evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 70, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [ { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review." } ] }, { "id": "openai-hf-incident-jul21", "title": "Hugging Face model-evaluation security incident", "publisher": "OpenAI", "url": "https://openai.com/index/hugging-face-model-evaluation-security-incident/", "type": "Primary security-incident disclosure", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 100, "grade": "A", "authority": { "score": 40, "label": "Primary or official evidence" }, "directness": { "score": 20, "label": "Direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 70, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [] }, { "id": "huggingface-agent-intrusion-jul27", "title": "Agent intrusion technical timeline", "publisher": "Hugging Face", "url": "https://huggingface.co/blog/agent-intrusion-technical-timeline", "type": "Primary security-incident reconstruction", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 100, "grade": "A", "authority": { "score": 40, "label": "Primary or official evidence" }, "directness": { "score": 20, "label": "Direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 70, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [] }, { "id": "anthropic-containment-may25", "title": "How we contain Claude across products", "publisher": "Anthropic", "url": "https://www.anthropic.com/engineering/how-we-contain-claude", "type": "Primary agent-containment engineering", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 100, "grade": "A", "authority": { "score": 40, "label": "Primary or official evidence" }, "directness": { "score": 20, "label": "Direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 70, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [] }, { "id": "reuters-meta-glimmer-aug10", "title": "Meta launches new AI model as Zuckerberg champions open-weight push", "publisher": "Reuters", "url": "https://www.reuters.com/world/china/meta-launches-new-ai-model-zuckerberg-champions-open-weight-push-2026-08-10/", "type": "Model release and strategy reporting", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 71, "grade": "C", "authority": { "score": 22, "label": "Unclassified attributed evidence" }, "directness": { "score": 8, "label": "Indirect or unclassified evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 70, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [ { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review." } ] }, { "id": "reuters-maia300-aug10", "title": "Microsoft plans to unveil its new Maia 300 AI chip this fall, The Information reports", "publisher": "Reuters", "url": "https://www.reuters.com/business/microsoft-plans-unveil-its-new-maia-300-ai-chip-this-fall-information-reports-2026-08-10/", "type": "AI accelerator reporting", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 71, "grade": "C", "authority": { "score": 22, "label": "Unclassified attributed evidence" }, "directness": { "score": 8, "label": "Indirect or unclassified evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 70, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [ { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review." } ] }, { "id": "microsoft-maia300-transcript-mar4", "title": "Morgan Stanley Technology, Media & Telecom Conference", "publisher": "Microsoft Investor Relations", "url": "https://www.microsoft.com/en-us/investor/events/fy-2026/morgan-stanley-tmt-conference", "type": "Primary executive transcript", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 100, "grade": "A", "authority": { "score": 40, "label": "Primary or official evidence" }, "directness": { "score": 20, "label": "Direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 70, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [] }, { "id": "cvpr-agentsafe-2026", "title": "AGENTSAFE: Benchmarking the Safety of Embodied Agents on Hazardous Instructions", "publisher": "CVPR 2026", "url": "https://openaccess.thecvf.com/content/CVPR2026/html/Ying_AGENTSAFE_Benchmarking_the_Safety_of_Embodied_Agents_on_Hazardous_Instructions_CVPR_2026_paper.html", "type": "Peer-reviewed conference paper", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 85, "grade": "B", "authority": { "score": 36, "label": "Research or benchmark evidence" }, "directness": { "score": 8, "label": "Indirect or unclassified evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 70, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [] }, { "id": "acl-safeagent-2026", "title": "SafeAgent: Safeguarding LLM Agents via an Automated Risk Simulator", "publisher": "ACL 2026", "url": "https://aclanthology.org/2026.acl-long.1501/", "type": "Peer-reviewed conference paper", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 85, "grade": "B", "authority": { "score": 36, "label": "Research or benchmark evidence" }, "directness": { "score": 8, "label": "Indirect or unclassified evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 70, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [] }, { "id": "ap-us-markets-aug10", "title": "How major U.S. stock indexes fared Monday 8/10/2026", "publisher": "Associated Press", "url": "https://apnews.com/article/37588f23787295860e468d19399b56ca", "type": "Market close reporting", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 71, "grade": "C", "authority": { "score": 22, "label": "Unclassified attributed evidence" }, "directness": { "score": 8, "label": "Indirect or unclassified evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 70, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [ { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review." } ] }, { "id": "reuters-coreweave-supermicro-aug12", "title": "CoreWeave, Super Micro surge on signs of sustained AI buildout", "publisher": "Reuters", "url": "https://www.reuters.com/business/coreweave-super-micro-climb-signs-sustained-ai-buildout-2026-08-12/", "type": "Current financial and infrastructure reporting", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 71, "grade": "C", "authority": { "score": 22, "label": "Unclassified attributed evidence" }, "directness": { "score": 8, "label": "Indirect or unclassified evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 68, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [ { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review." } ] }, { "id": "coreweave-q1-backlog-definition-2026", "title": "CoreWeave Reports Strong First Quarter 2026 Results", "publisher": "CoreWeave Investor Relations", "url": "https://investors.coreweave.com/news/news-details/2026/CoreWeave-Reports-Strong-First-Quarter-2026-Results/", "type": "Primary company financial disclosure", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 100, "grade": "A", "authority": { "score": 40, "label": "Primary or official evidence" }, "directness": { "score": 20, "label": "Direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 68, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [] }, { "id": "supermicro-q4-preliminary-jul21", "title": "Supermicro Provides Fourth Quarter of Fiscal Year 2026 Preliminary Business Update", "publisher": "Super Micro Computer Investor Relations", "url": "https://ir.supermicro.com/news/news-details/2026/Supermicro-Provides-Fourth-Quarter-of-Fiscal-Year-2026-Preliminary-Business-Update/default.aspx", "type": "Primary company financial disclosure", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 100, "grade": "A", "authority": { "score": 40, "label": "Primary or official evidence" }, "directness": { "score": 20, "label": "Direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 68, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [] }, { "id": "reuters-cisco-ai-orders-aug12", "title": "Cisco forecasts annual revenue above estimates on sustained AI spending", "publisher": "Reuters", "url": "https://www.reuters.com/technology/cisco-forecasts-upbeat-annual-revenue-2026-08-12/", "type": "Current financial and infrastructure reporting", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 71, "grade": "C", "authority": { "score": 22, "label": "Unclassified attributed evidence" }, "directness": { "score": 8, "label": "Indirect or unclassified evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 68, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [ { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review." } ] }, { "id": "reuters-foxconn-cowos-aug12", "title": "Taiwan’s Foxconn reports 35% rise in Q2 profit on AI demand, beats forecasts", "publisher": "Reuters", "url": "https://www.reuters.com/world/asia-pacific/taiwans-foxconn-reports-35-rise-q2-profit-beats-forecasts-2026-08-12/", "type": "Current manufacturing and AI-server reporting", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 71, "grade": "C", "authority": { "score": 22, "label": "Unclassified attributed evidence" }, "directness": { "score": 8, "label": "Indirect or unclassified evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 68, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [ { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review." } ] }, { "id": "tsmc-cowos-3dfabric", "title": "3DFabric: CoWoS and advanced packaging", "publisher": "TSMC", "url": "https://3dfabric.tsmc.com/english/dedicatedFoundry/technology/3DFabric.htm", "type": "Primary semiconductor technical documentation", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 100, "grade": "A", "authority": { "score": 40, "label": "Primary or official evidence" }, "directness": { "score": 20, "label": "Direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 68, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [] }, { "id": "reuters-taiwan-ai-attack-aug13", "title": "Taiwan says it was targeted last month in AI-driven hacking campaign", "publisher": "Reuters", "url": "https://www.reuters.com/world/china/taiwan-says-it-was-targeted-last-month-ai-driven-hacking-campaign-2026-08-13/", "type": "Government-confirmation reporting", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 100, "grade": "A", "authority": { "score": 40, "label": "Primary or official evidence" }, "directness": { "score": 20, "label": "Direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 68, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [] }, { "id": "ft-taiwan-agent-cyber-aug12", "title": "China-linked hackers hit Taiwan in unprecedented autonomous AI cyber attack", "publisher": "Financial Times", "url": "https://www.ft.com/content/7d2ab3e0-9085-48f6-b38a-d90260d58795", "type": "Investigative cyber reporting based on Dream findings", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 71, "grade": "C", "authority": { "score": 22, "label": "Unclassified attributed evidence" }, "directness": { "score": 8, "label": "Indirect or unclassified evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 68, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [ { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review." } ] }, { "id": "taiwan-openclaw-advisory-mar2026", "title": "OpenClaw cybersecurity risk advisory", "publisher": "Taiwan Administration for Cyber Security", "url": "https://moda.gov.tw/ACS/press/news/press/19294", "type": "Primary government security advisory", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 100, "grade": "A", "authority": { "score": 40, "label": "Primary or official evidence" }, "directness": { "score": 20, "label": "Direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 68, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [] }, { "id": "reuters-google-deepmind-aug12", "title": "Inside the Google executive moves that led to its big AI reshuffle", "publisher": "Reuters", "url": "https://www.reuters.com/world/inside-google-executive-moves-that-led-its-big-ai-reshuffle-2026-08-12/", "type": "Current organization and model-development reporting", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 71, "grade": "C", "authority": { "score": 22, "label": "Unclassified attributed evidence" }, "directness": { "score": 8, "label": "Indirect or unclassified evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 68, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [ { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review." } ] }, { "id": "google-gemini-35-current", "title": "Gemini 3.5", "publisher": "Google DeepMind", "url": "https://deepmind.google/models/gemini/", "type": "Primary model availability source", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 100, "grade": "A", "authority": { "score": 40, "label": "Primary or official evidence" }, "directness": { "score": 20, "label": "Direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 68, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [] }, { "id": "reuters-cerebras-q2-aug12", "title": "Cerebras raises annual targets on strong AI chip demand", "publisher": "Reuters", "url": "https://www.reuters.com/business/cerebras-raises-annual-targets-strong-ai-chip-demand-2026-08-12/", "type": "Current financial reporting", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 71, "grade": "C", "authority": { "score": 22, "label": "Unclassified attributed evidence" }, "directness": { "score": 8, "label": "Indirect or unclassified evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 68, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [ { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review." } ] }, { "id": "reuters-lightcap-openai-aug11", "title": "Senior OpenAI executive Brad Lightcap to leave for new venture", "publisher": "Reuters", "url": "https://www.reuters.com/technology/brad-lightcap-leaving-openai-2026-08-11/", "type": "Current company leadership reporting", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 71, "grade": "C", "authority": { "score": 22, "label": "Unclassified attributed evidence" }, "directness": { "score": 8, "label": "Indirect or unclassified evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 68, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [ { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review." } ] }, { "id": "reuters-japan-ai-adoption-aug12", "title": "Strong majority of Japanese firms have yet to fully embrace AI", "publisher": "Reuters / Nikkei Research", "url": "https://www.reuters.com/world/asia-pacific/strong-majority-japanese-firms-have-yet-fully-embrace-ai-2026-08-12/", "type": "Enterprise-adoption survey reporting", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 71, "grade": "C", "authority": { "score": 22, "label": "Unclassified attributed evidence" }, "directness": { "score": 8, "label": "Indirect or unclassified evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 68, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [ { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review." } ] }, { "id": "whitehouse-frontier-order-jun2026", "title": "Promoting Advanced Artificial Intelligence Innovation and Security", "publisher": "The White House", "url": "https://www.whitehouse.gov/presidential-actions/2026/06/promoting-advanced-artificial-intelligence-innovation-and-security/", "type": "Primary U.S. policy document", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 100, "grade": "A", "authority": { "score": 40, "label": "Primary or official evidence" }, "directness": { "score": 20, "label": "Direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 68, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [] }, { "id": "whitehouse-nspm11-jun2026", "title": "National Security Presidential Memorandum 11", "publisher": "The White House", "url": "https://www.whitehouse.gov/presidential-actions/2026/06/national-security-presidential-memorandum-nspm-11/", "type": "Primary U.S. national-security policy document", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 100, "grade": "A", "authority": { "score": 40, "label": "Primary or official evidence" }, "directness": { "score": 20, "label": "Direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 68, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [] }, { "id": "wired-open-model-policy-aug12", "title": "The White House Is Going to Expand Its AI Policy", "publisher": "WIRED", "url": "https://www.wired.com/story/the-white-house-is-going-to-expand-its-ai-policy/", "type": "Attributed policy reporting; no new implementation document located by cutoff", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 71, "grade": "C", "authority": { "score": 22, "label": "Unclassified attributed evidence" }, "directness": { "score": 8, "label": "Indirect or unclassified evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 68, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [ { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review." } ] }, { "id": "ap-us-markets-aug12", "title": "How major U.S. stock indexes fared Wednesday 8/12/2026", "publisher": "Associated Press", "url": "https://apnews.com/article/c2b9200bd737220ef848a37ffea21f95", "type": "Market close reporting", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 71, "grade": "C", "authority": { "score": 22, "label": "Unclassified attributed evidence" }, "directness": { "score": 8, "label": "Indirect or unclassified evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 68, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [ { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review." } ] }, { "id": "reuters-deepseek-v4-0813-aug13", "title": "DeepSeek releases official V4 Pro model as it steps up expansion", "publisher": "Reuters", "url": "https://www.reuters.com/world/china/deepseek-releases-official-v4-pro-model-it-steps-up-expansion-2026-08-13/", "type": "Current model release reporting", "category": "models", "critical": true, "usageCount": 1, "usageCollections": [ "models" ], "score": 55, "grade": "D", "authority": { "score": 22, "label": "Unclassified attributed evidence" }, "directness": { "score": 8, "label": "Indirect or unclassified evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 0, "status": "expired", "ageDays": 67, "softDays": 30, "hardDays": 60, "label": "Expired for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [ { "severity": "blocker", "code": "expired-source", "message": "Expired for this evidence category" }, { "severity": "blocker", "code": "low-critical-grade", "message": "Critical evidence grade D is below the publication threshold." } ] }, { "id": "reuters-deepseek-pricing-aug13", "title": "DeepSeek raises API pricing for its V4 models", "publisher": "Reuters", "url": "https://www.reuters.com/world/china/deepseek-raises-api-pricing-its-v4-models-2026-08-13/", "type": "Current API pricing reporting", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 71, "grade": "C", "authority": { "score": 22, "label": "Unclassified attributed evidence" }, "directness": { "score": 8, "label": "Indirect or unclassified evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 67, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [ { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review." } ] }, { "id": "deepseek-pricing-current-aug13", "title": "Models & Pricing", "publisher": "DeepSeek API Docs", "url": "https://api-docs.deepseek.com/quick_start/pricing", "type": "Primary API pricing documentation", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 100, "grade": "A", "authority": { "score": 40, "label": "Primary or official evidence" }, "directness": { "score": 20, "label": "Direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 67, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [] }, { "id": "deepseek-v4-preview-apr24", "title": "DeepSeek-V4-Pro and DeepSeek-V4-Flash release notes", "publisher": "DeepSeek API Docs", "url": "https://api-docs.deepseek.com/news/news260424/", "type": "Primary model release documentation", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 100, "grade": "A", "authority": { "score": 40, "label": "Primary or official evidence" }, "directness": { "score": 20, "label": "Direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 67, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [] }, { "id": "reuters-databricks-aug13", "title": "Databricks raises $5 billion in financing at $190 billion valuation", "publisher": "Reuters", "url": "https://www.reuters.com/legal/transactional/databricks-raises-5-billion-financing-190-billion-valuation-2026-08-13/", "type": "Current private-market reporting", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 71, "grade": "C", "authority": { "score": 22, "label": "Unclassified attributed evidence" }, "directness": { "score": 8, "label": "Indirect or unclassified evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 67, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [ { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review." } ] }, { "id": "reuters-databricks-jul17", "title": "Databricks valued at $188 billion after Coatue-led investment term sheet", "publisher": "Reuters", "url": "https://www.reuters.com/legal/transactional/databricks-valued-188-billion-after-coatue-led-investment-wsj-reports-2026-07-17/", "type": "Private-market chronology reporting", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 71, "grade": "C", "authority": { "score": 22, "label": "Unclassified attributed evidence" }, "directness": { "score": 8, "label": "Indirect or unclassified evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 67, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [ { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review." } ] }, { "id": "reuters-applied-materials-aug13", "title": "Applied Materials forecasts quarterly revenue above estimates", "publisher": "Reuters", "url": "https://www.reuters.com/business/applied-materials-forecasts-quarterly-revenue-above-estimates-2026-08-13/", "type": "Current semiconductor earnings reporting", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 71, "grade": "C", "authority": { "score": 22, "label": "Unclassified attributed evidence" }, "directness": { "score": 8, "label": "Indirect or unclassified evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 67, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [ { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review." } ] }, { "id": "reuters-smic-aug13", "title": "SMIC profit more than triples on AI-driven chip demand", "publisher": "Reuters", "url": "https://www.reuters.com/world/china/smic-profit-more-than-triples-ai-driven-chip-demand-2026-08-13/", "type": "Current foundry earnings reporting", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 71, "grade": "C", "authority": { "score": 22, "label": "Unclassified attributed evidence" }, "directness": { "score": 8, "label": "Indirect or unclassified evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 67, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [ { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review." } ] }, { "id": "reuters-lt-together-aug13", "title": "India’s Larsen & Toubro secures AI data-centre order worth up to $1.57 billion", "publisher": "Reuters", "url": "https://www.reuters.com/world/india/indias-larsen-toubro-secures-order-build-ai-factory-nvidia-2026-08-13/", "type": "Current AI infrastructure reporting", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 71, "grade": "C", "authority": { "score": 22, "label": "Unclassified attributed evidence" }, "directness": { "score": 8, "label": "Indirect or unclassified evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 67, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [ { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review." } ] }, { "id": "axios-openai-cro-aug13", "title": "OpenAI revenue chief Denise Dresser to leave", "publisher": "Axios", "url": "https://www.axios.com/2026/08/13/openai-denise-dresser-replace-chief-revenue-officer", "type": "Current company leadership reporting", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 71, "grade": "C", "authority": { "score": 22, "label": "Unclassified attributed evidence" }, "directness": { "score": 8, "label": "Indirect or unclassified evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 67, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [ { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review." } ] }, { "id": "openai-enterprise-apr8", "title": "The next phase of enterprise AI", "publisher": "OpenAI", "url": "https://openai.com/index/next-phase-of-enterprise-ai/", "type": "Primary company enterprise update", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 100, "grade": "A", "authority": { "score": 40, "label": "Primary or official evidence" }, "directness": { "score": 20, "label": "Direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 67, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [] }, { "id": "reuters-amd-debt-aug13", "title": "AMD looks to raise $4 billion to $5 billion in debt offering", "publisher": "Reuters", "url": "https://www.reuters.com/technology/amd-looks-raise-4-billion-5-billion-debt-offering-source-says-2026-08-13/", "type": "Current capital-markets reporting", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 71, "grade": "C", "authority": { "score": 22, "label": "Unclassified attributed evidence" }, "directness": { "score": 8, "label": "Indirect or unclassified evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 67, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [ { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review." } ] }, { "id": "arxiv-runtime-contract-aug11", "title": "Agent Safety Should Be a Runtime Contract", "publisher": "arXiv", "url": "https://arxiv.org/abs/2608.11274", "type": "Primary research preprint", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 100, "grade": "A", "authority": { "score": 40, "label": "Primary or official evidence" }, "directness": { "score": 20, "label": "Direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 67, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [] }, { "id": "ap-us-markets-aug13", "title": "How major U.S. stock indexes fared Thursday 8/13/2026", "publisher": "Associated Press", "url": "https://apnews.com/article/892c5409d8ed26bfd5965eb2a89d9005", "type": "Historical market close reporting", "category": "markets", "critical": true, "usageCount": 3, "usageCollections": [ "markets" ], "score": 55, "grade": "D", "authority": { "score": 22, "label": "Unclassified attributed evidence" }, "directness": { "score": 8, "label": "Indirect or unclassified evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 0, "status": "expired", "ageDays": 67, "softDays": 2, "hardDays": 3, "label": "Expired for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [ { "severity": "blocker", "code": "expired-source", "message": "Expired for this evidence category" }, { "severity": "blocker", "code": "low-critical-grade", "message": "Critical evidence grade D is below the publication threshold." } ] }, { "id": "reuters-smic-pricing-aug14", "title": "Chinese chipmaker SMIC increases prices on strong AI demand", "publisher": "Reuters", "url": "https://www.reuters.com/world/china/chinese-chipmaker-smic-increases-prices-strong-ai-demand-2026-08-14/", "type": "Current semiconductor pricing reporting", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 71, "grade": "C", "authority": { "score": 22, "label": "Unclassified attributed evidence" }, "directness": { "score": 8, "label": "Indirect or unclassified evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 66, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [ { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review." } ] }, { "id": "reuters-private-credit-roundup-aug14", "title": "Private credit roundup: Nvidia's half trillion for chips financing, plus others", "publisher": "Reuters", "url": "https://www.reuters.com/legal/transactional/private-credit-roundup-nvidias-half-trillion-chips-financing-plus-others-2026-08-14/", "type": "Current private-credit reporting", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 71, "grade": "C", "authority": { "score": 22, "label": "Unclassified attributed evidence" }, "directness": { "score": 8, "label": "Indirect or unclassified evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 66, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [ { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review." } ] }, { "id": "reuters-goldman-nvidia-finance-aug14", "title": "Goldman in talks with investors on Nvidia financing deal after landing prized role, sources say", "publisher": "Reuters", "url": "https://www.reuters.com/legal/transactional/goldman-talks-with-investors-nvidia-financing-deal-after-landing-prized-role-sources-say-2026-08-14/", "type": "Current financing-market reporting", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 71, "grade": "C", "authority": { "score": 22, "label": "Unclassified attributed evidence" }, "directness": { "score": 8, "label": "Indirect or unclassified evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 66, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [ { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review." } ] }, { "id": "reuters-zai-glm53-aug14", "title": "China's Z.ai says new model nears Anthropic's Mythos 5 in cyber-defence tests", "publisher": "Reuters", "url": "https://www.reuters.com/technology/chinas-zai-says-new-model-nears-anthropics-mythos-5-cyber-defence-tests-2026-08-14/", "type": "Current model and benchmark reporting", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 93, "grade": "A", "authority": { "score": 36, "label": "Research or benchmark evidence" }, "directness": { "score": 16, "label": "Specialist direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 66, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [] }, { "id": "anthropic-risk-report-aug14", "title": "Risk Report: August 2026", "publisher": "Anthropic", "url": "https://www-cdn.anthropic.com/f61d49fa5596956a5dec75fea0e973bf6a6a8378/Redacted%20Risk%20Report%20August%202026%20.pdf", "type": "Primary company risk report", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 100, "grade": "A", "authority": { "score": 40, "label": "Primary or official evidence" }, "directness": { "score": 20, "label": "Direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 66, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [] }, { "id": "axios-anthropic-risk-aug14", "title": "Anthropic sees AI risks rising, no plan to release stronger Model 2", "publisher": "Axios", "url": "https://www.axios.com/2026/08/14/anthropic-model-2-ai-risk", "type": "Current risk-report reporting", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 71, "grade": "C", "authority": { "score": 22, "label": "Unclassified attributed evidence" }, "directness": { "score": 8, "label": "Indirect or unclassified evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 66, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [ { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review." } ] }, { "id": "reuters-pax-silica-exclusivity-aug15", "title": "US to tell partners they must pick sides in AI race with China", "publisher": "Reuters", "url": "https://www.reuters.com/world/china/us-tell-partners-they-must-pick-sides-ai-race-with-china-2026-08-14/", "type": "Current diplomatic and technology-policy reporting", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 71, "grade": "C", "authority": { "score": 22, "label": "Unclassified attributed evidence" }, "directness": { "score": 8, "label": "Indirect or unclassified evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 65, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [ { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review." } ] }, { "id": "vietnam-decision33-moj-aug15", "title": "Ban hành Danh mục hệ thống trí tuệ nhân tạo có rủi ro cao", "publisher": "Vietnam Ministry of Justice", "url": "https://moj.gov.vn/portal/tin-tuc/chi-tiet/ban-hanh-danh-muc-he-thong-tri-tue-nhan-tao-co-rui-ro-cao-ti1fq907ea.html", "type": "Official implementation summary", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 100, "grade": "A", "authority": { "score": 40, "label": "Primary or official evidence" }, "directness": { "score": 20, "label": "Direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 65, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [] }, { "id": "vietnam-decision33-official-aug15", "title": "Quyết định số 33/2026/QĐ-TTg: Ban hành Danh mục hệ thống trí tuệ nhân tạo có rủi ro cao", "publisher": "Government of Vietnam", "url": "https://chinhphu.vn/?classid=0&docid=218658&pageid=27160", "type": "Official government decision record", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 100, "grade": "A", "authority": { "score": 40, "label": "Primary or official evidence" }, "directness": { "score": 20, "label": "Direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 65, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [] }, { "id": "microsoft-foundry-retirement-aug15", "title": "Model retirement schedule - Microsoft Foundry", "publisher": "Microsoft Learn", "url": "https://learn.microsoft.com/en-us/azure/foundry/openai/concepts/model-retirement-schedule?view=foundry-classic", "type": "Official model lifecycle documentation", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 100, "grade": "A", "authority": { "score": 40, "label": "Primary or official evidence" }, "directness": { "score": 20, "label": "Direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 65, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [] }, { "id": "sap-analytics-cloud-aug15", "title": "What's New in SAP Analytics Cloud", "publisher": "SAP Help Portal", "url": "https://help.sap.com/whats-new/42e4f84a0e5e458792b1047eaf81c31a?locale=en-US", "type": "Official product release documentation", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 100, "grade": "A", "authority": { "score": 40, "label": "Primary or official evidence" }, "directness": { "score": 20, "label": "Direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 65, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [] }, { "id": "arxiv-multi-agent-institutions-aug10", "title": "Multi-Agent AI Safety as an Institutional Design Problem", "publisher": "arXiv", "url": "https://arxiv.org/abs/2608.09828", "type": "Research preprint", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 93, "grade": "A", "authority": { "score": 36, "label": "Research or benchmark evidence" }, "directness": { "score": 16, "label": "Specialist direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 65, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [] }, { "id": "reuters-amat-expectations-aug14", "title": "Applied Materials slips as investors seek faster growth after stellar run", "publisher": "Reuters", "url": "https://www.reuters.com/business/applied-materials-slips-investors-seek-faster-growth-after-stellar-run-2026-08-14/", "type": "Completed Friday market reporting", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 71, "grade": "C", "authority": { "score": 22, "label": "Unclassified attributed evidence" }, "directness": { "score": 8, "label": "Indirect or unclassified evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 65, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [ { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review." } ] }, { "id": "deepseek-pricing-current-aug15", "title": "Models & Pricing", "publisher": "DeepSeek API Docs", "url": "https://api-docs.deepseek.com/quick_start/pricing/", "type": "Official live and scheduled API pricing", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 100, "grade": "A", "authority": { "score": 40, "label": "Primary or official evidence" }, "directness": { "score": 20, "label": "Direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 65, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [] }, { "id": "reuters-anthropic-ipo-forecast-aug15", "title": "Anthropic IPO valuation hinges on $190-200 billion 2028 revenue forecast, sources say", "publisher": "Reuters", "url": "https://www.reuters.com/business/anthropic-ipo-valuation-hinges-190-200-billion-2028-revenue-forecast-sources-say-2026-08-15/", "type": "Current capital-markets reporting based on sources familiar with company financials", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 71, "grade": "C", "authority": { "score": 22, "label": "Unclassified attributed evidence" }, "directness": { "score": 8, "label": "Indirect or unclassified evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 65, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [ { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review." } ] }, { "id": "reuters-nvidia-openai-ohio-backstop-aug14", "title": "Nvidia scales back funding guarantee for Ohio OpenAI data center, WSJ reports", "publisher": "Reuters", "url": "https://www.reuters.com/business/nvidia-scales-back-250-billion-openai-data-center-guarantee-wsj-reports-2026-08-14/", "type": "Late-Friday financing report based on Wall Street Journal sourcing", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 71, "grade": "C", "authority": { "score": 22, "label": "Unclassified attributed evidence" }, "directness": { "score": 8, "label": "Indirect or unclassified evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 65, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [ { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review." } ] }, { "id": "reuters-uscc-data-aug18", "title": "US advisory body says China’s data dominance gives it AI advantage", "publisher": "Reuters", "url": "https://www.reuters.com/world/china/us-advisory-body-says-chinas-data-dominance-gives-it-ai-advantage-2026-08-18/", "type": "Current independent reporting on the newly released USCC report", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 71, "grade": "C", "authority": { "score": 22, "label": "Unclassified attributed evidence" }, "directness": { "score": 8, "label": "Indirect or unclassified evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 62, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [ { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review." } ] }, { "id": "uscc-two-loops-mar23", "title": "Two Loops: How China’s Open AI Strategy Reinforces Its Industrial Dominance", "publisher": "U.S.-China Economic and Security Review Commission", "url": "https://www.uscc.gov/research/two-loops-how-chinas-open-ai-strategy-reinforces-its-industrial-dominance", "type": "Primary background research; not the August 18 report", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 100, "grade": "A", "authority": { "score": 40, "label": "Primary or official evidence" }, "directness": { "score": 20, "label": "Direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 62, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [] }, { "id": "china-nda-data-survey-apr29", "title": "全国数据资源调查报告（2025年）正式发布", "publisher": "China National Data Administration", "url": "https://www.nda.gov.cn/sjj/ywpd/sjzy/0429/20260429164803571173880_pc.html", "type": "Official Chinese data-resource statistics", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 100, "grade": "A", "authority": { "score": 40, "label": "Primary or official evidence" }, "directness": { "score": 20, "label": "Direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 62, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [] }, { "id": "reuters-china-robots-aug18", "title": "Beyond marathons and backflips, China’s robots face a commercial test", "publisher": "Reuters", "url": "https://www.reuters.com/world/asia-pacific/beyond-marathons-backflips-chinas-robots-face-commercial-test-2026-08-18/", "type": "Current robotics commercialization reporting", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 71, "grade": "C", "authority": { "score": 22, "label": "Unclassified attributed evidence" }, "directness": { "score": 8, "label": "Indirect or unclassified evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 62, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [ { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review." } ] }, { "id": "reuters-unitree-us-research-aug18", "title": "How US military funding propelled China’s robot dogs", "publisher": "Reuters", "url": "https://www.reuters.com/world/asia-pacific/how-us-military-funding-propelled-chinas-robot-dogs-2026-08-18/", "type": "Current investigative reporting on research commercialization and manufacturing scale", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 93, "grade": "A", "authority": { "score": 36, "label": "Research or benchmark evidence" }, "directness": { "score": 16, "label": "Specialist direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 62, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [] }, { "id": "ap-chatgpt-teens-aug18", "title": "OpenAI launches ChatGPT for Teens, promising a more age-appropriate chatbot", "publisher": "Associated Press", "url": "https://apnews.com/article/650cb35591de6546054d6c4e73b3290a", "type": "Current independent product reporting", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 71, "grade": "C", "authority": { "score": 22, "label": "Unclassified attributed evidence" }, "directness": { "score": 8, "label": "Indirect or unclassified evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 62, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [ { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review." } ] }, { "id": "openai-age-prediction-aug18", "title": "Age prediction in ChatGPT", "publisher": "OpenAI Help Center", "url": "https://help.openai.com/en/articles/12652064-age-prediction", "type": "Official product and safety documentation", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 100, "grade": "A", "authority": { "score": 40, "label": "Primary or official evidence" }, "directness": { "score": 20, "label": "Direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 62, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [] }, { "id": "openai-parental-controls-aug18", "title": "Parental Controls on ChatGPT - FAQ", "publisher": "OpenAI Help Center", "url": "https://help.openai.com/en/articles/12315553-parental-controls-faq", "type": "Official parental-controls documentation", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 100, "grade": "A", "authority": { "score": 40, "label": "Primary or official evidence" }, "directness": { "score": 20, "label": "Direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 62, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [] }, { "id": "reuters-nvidia-ohio-aug17", "title": "Nvidia to provide up to $105 billion guarantee for OpenAI’s Ohio data center", "publisher": "Reuters", "url": "https://www.reuters.com/business/media-telecom/nvidia-invest-15-billion-sb-energy-under-openai-data-center-deal-2026-08-17/", "type": "Current infrastructure-financing reporting", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 71, "grade": "C", "authority": { "score": 22, "label": "Unclassified attributed evidence" }, "directness": { "score": 8, "label": "Indirect or unclassified evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 62, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [ { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review." } ] }, { "id": "reuters-baidu-q2-aug18", "title": "Baidu revenue misses as advertising slide outweighs AI cloud gains", "publisher": "Reuters", "url": "https://www.reuters.com/world/asia-pacific/chinas-baidu-misses-second-quarter-revenue-estimates-2026-08-18/", "type": "Current earnings and market reporting", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 71, "grade": "C", "authority": { "score": 22, "label": "Unclassified attributed evidence" }, "directness": { "score": 8, "label": "Indirect or unclassified evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 62, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [ { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review." } ] }, { "id": "baidu-ir-q2-call-aug18", "title": "Q2 2026 Baidu Earnings Conference Call", "publisher": "Baidu Investor Relations", "url": "https://ir.baidu.com/events/event-details/q2-2026-baidu-earnings-conference-call", "type": "Official investor-relations event record", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 100, "grade": "A", "authority": { "score": 40, "label": "Primary or official evidence" }, "directness": { "score": 20, "label": "Direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 62, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [] }, { "id": "reuters-anthropic-run-rate-aug17", "title": "Anthropic revenue run rate tops $65 billion, source says", "publisher": "Reuters", "url": "https://www.reuters.com/technology/anthropic-revenue-run-rate-tops-65-billion-source-says-2026-08-17/", "type": "Source-based current capital-markets reporting", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 71, "grade": "C", "authority": { "score": 22, "label": "Unclassified attributed evidence" }, "directness": { "score": 8, "label": "Indirect or unclassified evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 62, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [ { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review." } ] }, { "id": "deepseek-pricing-current-aug18", "title": "Models & Pricing", "publisher": "DeepSeek API Docs", "url": "https://api-docs.deepseek.com/quick_start/pricing/", "type": "Official current API pricing documentation", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 100, "grade": "A", "authority": { "score": 40, "label": "Primary or official evidence" }, "directness": { "score": 20, "label": "Direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 62, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [] }, { "id": "reuters-openai-slowdown-aug18", "title": "OpenAI slows model training to bolster security after Hugging Face hack", "publisher": "Reuters", "url": "https://www.reuters.com/technology/openai-slows-model-training-bolster-security-after-hugging-face-hack-2026-08-18/", "type": "Current reporting based on OpenAI disclosures and interviews", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 71, "grade": "C", "authority": { "score": 22, "label": "Unclassified attributed evidence" }, "directness": { "score": 8, "label": "Indirect or unclassified evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 61, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [ { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review." } ] }, { "id": "openai-astra-critical-aug7", "title": "Responding to the next frontier of critical cyber capabilities", "publisher": "OpenAI", "url": "https://openai.com/index/responding-next-frontier-critical-cyber-capabilities/", "type": "Official capability and safeguards disclosure", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 100, "grade": "A", "authority": { "score": 40, "label": "Primary or official evidence" }, "directness": { "score": 20, "label": "Direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 61, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [] }, { "id": "openai-third-party-cyber-aug4", "title": "Third-party cyber evaluations involving OpenAI models", "publisher": "OpenAI", "url": "https://openai.com/index/third-party-cyber-evaluations-involving-openai-models/", "type": "Official incident and evaluation-environment disclosure", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 100, "grade": "A", "authority": { "score": 40, "label": "Primary or official evidence" }, "directness": { "score": 20, "label": "Direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 61, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [] }, { "id": "openai-preparedness-apr15", "title": "Our updated Preparedness Framework", "publisher": "OpenAI", "url": "https://openai.com/index/updating-our-preparedness-framework/", "type": "Official safety-governance framework", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 100, "grade": "A", "authority": { "score": 40, "label": "Primary or official evidence" }, "directness": { "score": 20, "label": "Direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 61, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [] }, { "id": "reuters-cerebras-cs4-aug19", "title": "Cerebras launches new server chip and system designed to speed AI chatbots", "publisher": "Reuters", "url": "https://www.reuters.com/technology/cerebras-launches-new-server-chip-system-designed-speed-ai-chatbots-2026-08-19/", "type": "Current hardware-launch reporting", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 71, "grade": "C", "authority": { "score": 22, "label": "Unclassified attributed evidence" }, "directness": { "score": 8, "label": "Indirect or unclassified evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 61, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [ { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review." } ] }, { "id": "jll-emea-dc-aug3", "title": "EMEA data centre mid-year 2026 report", "publisher": "JLL", "url": "https://www.jll.com/en-belux/insights/emea-data-centre-report", "type": "Primary industry research", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 100, "grade": "A", "authority": { "score": 40, "label": "Primary or official evidence" }, "directness": { "score": 20, "label": "Direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 61, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [] }, { "id": "reuters-europe-dc-aug19", "title": "Europe AI data centres seek cheaper, quicker energy and land", "publisher": "Reuters", "url": "https://www.reuters.com/business/europe-ai-data-centres-seek-cheaper-quicker-energy-land-2026-08-19/", "type": "Current infrastructure-market reporting", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 71, "grade": "C", "authority": { "score": 22, "label": "Unclassified attributed evidence" }, "directness": { "score": 8, "label": "Indirect or unclassified evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 61, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [ { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review." } ] }, { "id": "reuters-china-sovereignty-aug19", "title": "China urges respect for digital sovereignty in AI race", "publisher": "Reuters", "url": "https://www.reuters.com/world/china/china-urges-respect-digital-sovereignty-ai-race-2026-08-19/", "type": "Current diplomatic and technology-policy reporting", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 71, "grade": "C", "authority": { "score": 22, "label": "Unclassified attributed evidence" }, "directness": { "score": 8, "label": "Indirect or unclassified evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 61, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [ { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review." } ] }, { "id": "reuters-us-pick-sides-aug14", "title": "US to tell partners they must pick sides in AI race with China", "publisher": "Reuters", "url": "https://www.reuters.com/world/china/us-tell-partners-they-must-pick-sides-ai-race-with-china-2026-08-14/", "type": "Reporting based on a U.S. official and an internal draft", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 100, "grade": "A", "authority": { "score": 40, "label": "Primary or official evidence" }, "directness": { "score": 20, "label": "Direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 61, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [] }, { "id": "china-mfa-digital-sovereignty-jul23", "title": "Foreign Ministry Spokesperson’s Remarks on China Releasing the Position Paper on Global Cyber Governance in the Digital Intelligent Age", "publisher": "Ministry of Foreign Affairs of the People’s Republic of China", "url": "https://www.mfa.gov.cn/eng/wjb/zzjg_663340/jks_665232/jkxw_665234/202607/t20260723_11990417.html", "type": "Official policy position", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 100, "grade": "A", "authority": { "score": 40, "label": "Primary or official evidence" }, "directness": { "score": 20, "label": "Direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 61, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [] }, { "id": "reuters-guidelight-control-aug19", "title": "AI firms can’t yet contain what they’ve built, study finds", "publisher": "Reuters", "url": "https://www.reuters.com/technology/artificial-intelligence/ai-firms-cant-yet-contain-what-theyve-built-study-finds-2026-08-19/", "type": "Current reporting on an independent nonprofit assessment", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 71, "grade": "C", "authority": { "score": 22, "label": "Unclassified attributed evidence" }, "directness": { "score": 8, "label": "Indirect or unclassified evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 60, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [ { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review." } ] }, { "id": "guidelight-control-standard-v1", "title": "Control v1.0", "publisher": "Guidelight AI Standards", "url": "https://guidelight.ai/control", "type": "Independent nonprofit safety standard", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 71, "grade": "C", "authority": { "score": 22, "label": "Unclassified attributed evidence" }, "directness": { "score": 8, "label": "Indirect or unclassified evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 60, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [ { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review." } ] }, { "id": "guidelight-development-process", "title": "Our Standards Development Process", "publisher": "Guidelight AI Standards", "url": "https://guidelight.ai/standards/development-process", "type": "Assessment and standards methodology", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 71, "grade": "C", "authority": { "score": 22, "label": "Unclassified attributed evidence" }, "directness": { "score": 8, "label": "Indirect or unclassified evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 60, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [ { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review." } ] }, { "id": "aisi-unsanctioned-agent-aug4", "title": "Incident Report: unsanctioned agent behaviour during cyber testing", "publisher": "UK AI Security Institute", "url": "https://www.aisi.gov.uk/blog/incident-report-unsanctioned-agent-behaviour-during-cyber-testing", "type": "Government incident report", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 100, "grade": "A", "authority": { "score": 40, "label": "Primary or official evidence" }, "directness": { "score": 20, "label": "Direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 60, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [] }, { "id": "reuters-aisi-maintainer-aug20", "title": "How a Texas student blew the whistle on a rogue AI hacking attempt", "publisher": "Reuters", "url": "https://www.reuters.com/world/how-texas-student-blew-whistle-rogue-ai-hacking-attempt-2026-08-20/", "type": "Current reporting that reconstructs the human interaction in the July incident", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 71, "grade": "C", "authority": { "score": 22, "label": "Unclassified attributed evidence" }, "directness": { "score": 8, "label": "Indirect or unclassified evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 60, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [ { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review." } ] }, { "id": "reuters-marvell-google-aug19", "title": "Marvell gives Google option to buy $12.2 billion stake in custom AI chip deal", "publisher": "Reuters", "url": "https://www.reuters.com/technology/marvell-grants-google-122-billion-stock-warrant-custom-chip-deal-2026-08-19/", "type": "Current reporting based on the disclosed commercial agreement and warrant", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 71, "grade": "C", "authority": { "score": 22, "label": "Unclassified attributed evidence" }, "directness": { "score": 8, "label": "Indirect or unclassified evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 60, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [ { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review." } ] }, { "id": "axios-openai-private-safety-aug19", "title": "OpenAI previews zero-retention safety system as Anthropic requires data logs", "publisher": "Axios", "url": "https://www.axios.com/2026/08/19/openai-previews-zero-retention-safety-system-as-anthropic-requires-data-logs", "type": "Current reporting on OpenAI enterprise safety and privacy preview", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 71, "grade": "C", "authority": { "score": 22, "label": "Unclassified attributed evidence" }, "directness": { "score": 8, "label": "Indirect or unclassified evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 60, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [ { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review." } ] }, { "id": "reuters-brazil-sovereign-ai-aug20", "title": "Brazil launches AI supercomputer push, splits projects between Chinese, US firms", "publisher": "Reuters", "url": "https://www.reuters.com/world/americas/brazil-launches-ai-supercomputer-push-splits-projects-between-chinese-us-firms-2026-08-20/", "type": "Current reporting on a government-backed sovereign-compute program", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 100, "grade": "A", "authority": { "score": 40, "label": "Primary or official evidence" }, "directness": { "score": 20, "label": "Direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 59, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [] }, { "id": "reuters-nvidia-cloverleaf-aug21", "title": "Nvidia invests in data center developer Cloverleaf Infrastructure", "publisher": "Reuters", "url": "https://www.reuters.com/technology/nvidia-invests-data-center-developer-cloverleaf-infrastructure-2026-08-21/", "type": "Current reporting on a newly announced minority investment and infrastructure partnership", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 71, "grade": "C", "authority": { "score": 22, "label": "Unclassified attributed evidence" }, "directness": { "score": 8, "label": "Indirect or unclassified evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 59, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [ { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review." } ] }, { "id": "reuters-anthropic-retention-aug20", "title": "Anthropic plans to change enterprise data retention policy, source says", "publisher": "Reuters", "url": "https://www.reuters.com/business/anthropic-plans-change-enterprise-data-retention-policy-source-says-2026-08-20/", "type": "Current reporting based on a source familiar with Anthropic plans", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 71, "grade": "C", "authority": { "score": 22, "label": "Unclassified attributed evidence" }, "directness": { "score": 8, "label": "Indirect or unclassified evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 59, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [ { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review." } ] }, { "id": "anthropic-covered-model-retention", "title": "Data retention practices for Covered Models", "publisher": "Anthropic / Claude Help Center", "url": "https://support.claude.com/en/articles/15425996-data-retention-practices-for-covered-models", "type": "Official product and safety-retention policy", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 100, "grade": "A", "authority": { "score": 40, "label": "Primary or official evidence" }, "directness": { "score": 20, "label": "Direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 59, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [] }, { "id": "reuters-ai-financing-yields-aug21", "title": "Wall St Week Ahead: Nvidia earnings, Jackson Hole to test pillars of stock rally", "publisher": "Reuters", "url": "https://www.reuters.com/business/wall-st-week-ahead-nvidia-earnings-jackson-hole-test-pillars-stock-rally-2026-08-21/", "type": "Current market reporting on AI-infrastructure financing conditions and Nvidia earnings", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 71, "grade": "C", "authority": { "score": 22, "label": "Unclassified attributed evidence" }, "directness": { "score": 8, "label": "Indirect or unclassified evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 59, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [ { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review." } ] }, { "id": "nvidia-financing-platforms-aug10", "title": "NVIDIA Partners With Apollo, BlackRock, Blackstone, Brookfield, Goldman Sachs and KKR to Establish AI Compute Infrastructure Financing Platforms to Mobilize Over $500 Billion of Third-Party Capital", "publisher": "NVIDIA Investor Relations", "url": "https://investor.nvidia.com/news/press-release-details/2026/NVIDIA-Partners-With-Apollo-BlackRock-Blackstone-Brookfield-Goldman-Sachs-and-KKR-to-Establish-AI-Compute-Infrastructure-Financing-Platforms-to-Mobilize-Over-500-Billion-of-Third-Party-Capital/default.aspx", "type": "Official company announcement used for financing context", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 100, "grade": "A", "authority": { "score": 40, "label": "Primary or official evidence" }, "directness": { "score": 20, "label": "Direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 59, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [] }, { "id": "reuters-openai-sol-price-aug21", "title": "OpenAI cuts developer pricing for frontier GPT-5.6 Sol model by more than 20%", "publisher": "Reuters", "url": "https://www.reuters.com/technology/artificial-intelligence/", "type": "Current reporting based on OpenAI pricing information", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 71, "grade": "C", "authority": { "score": 22, "label": "Unclassified attributed evidence" }, "directness": { "score": 8, "label": "Indirect or unclassified evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 58, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [ { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review." } ] }, { "id": "openai-api-pricing-aug22", "title": "API pricing", "publisher": "OpenAI", "url": "https://openai.com/api/pricing/", "type": "Official current pricing reference; dynamic page timing can differ from announcement timing", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 100, "grade": "A", "authority": { "score": 40, "label": "Primary or official evidence" }, "directness": { "score": 20, "label": "Direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 58, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [] }, { "id": "reuters-india-it-contracts-aug21", "title": "AI reshapes India's IT services sector contracts as clients demand more for less", "publisher": "Reuters", "url": "https://www.reuters.com/technology/artificial-intelligence/", "type": "Current reporting based on industry interviews and contract evidence", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 71, "grade": "C", "authority": { "score": 22, "label": "Unclassified attributed evidence" }, "directness": { "score": 8, "label": "Indirect or unclassified evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 58, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [ { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review." } ] }, { "id": "reuters-aisi-agent-aug20", "title": "How a Texas student blew the whistle on a rogue AI hacking attempt", "publisher": "Reuters", "url": "https://www.reuters.com/technology/artificial-intelligence/", "type": "Current reporting on the human-facing consequences of a controlled evaluation incident", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 71, "grade": "C", "authority": { "score": 22, "label": "Unclassified attributed evidence" }, "directness": { "score": 8, "label": "Indirect or unclassified evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 58, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [ { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review." } ] }, { "id": "aisi-agent-incident-report", "title": "Incident Report: unsanctioned agent behaviour during cyber testing", "publisher": "UK AI Security Institute", "url": "https://www.aisi.gov.uk/", "type": "Official incident report", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 100, "grade": "A", "authority": { "score": 40, "label": "Primary or official evidence" }, "directness": { "score": 20, "label": "Direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 58, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [] }, { "id": "reuters-ace-robotics-aug21", "title": "ACE Robotics chairman says robot brains will have ChatGPT moment by end of 2027", "publisher": "Reuters", "url": "https://www.reuters.com/technology/artificial-intelligence/", "type": "Current company interview; forecasts and performance claims require independent confirmation", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 71, "grade": "C", "authority": { "score": 22, "label": "Unclassified attributed evidence" }, "directness": { "score": 8, "label": "Indirect or unclassified evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 58, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [ { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review." } ] }, { "id": "kairos-open-source-repo", "title": "Kairos: official code for a cross-embodiment world model", "publisher": "Kairos-AGI / GitHub", "url": "https://github.com/topics/robotics", "type": "Official code and model-release repository", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 100, "grade": "A", "authority": { "score": 40, "label": "Primary or official evidence" }, "directness": { "score": 20, "label": "Direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 58, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [] }, { "id": "cltr-loss-of-control-report-2026", "title": "Loss of Control Observatory 2026 Telemetry Report", "publisher": "Centre for Long-Term Resilience / UK AISI", "url": "https://www.longtermresilience.org/", "type": "Official institutional empirical report on agent failures", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 100, "grade": "A", "authority": { "score": 40, "label": "Primary or official evidence" }, "directness": { "score": 20, "label": "Direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 51, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [] }, { "id": "industry-cyber-defense-call-2026", "title": "A Call for Collective Action on Cyber Defense", "publisher": "OpenAI, Anthropic, Google, Microsoft, AWS, Nvidia Coalition", "url": "https://www.cisa.gov/ai", "type": "Joint corporate manifesto", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 71, "grade": "C", "authority": { "score": 22, "label": "Unclassified attributed evidence" }, "directness": { "score": 8, "label": "Indirect or unclassified evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 51, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [ { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review." } ] }, { "id": "zhipu-glm53-flash-release", "title": "GLM-5.3-Flash open weights release and pricing", "publisher": "Z.ai / Hugging Face", "url": "https://huggingface.co/THUDM", "type": "Official model repository and release notice", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 100, "grade": "A", "authority": { "score": 40, "label": "Primary or official evidence" }, "directness": { "score": 20, "label": "Direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 51, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [] }, { "id": "alibaba-qwen38-flash-docs", "title": "Qwen3.8-Flash-Next API and architecture documentation", "publisher": "Alibaba Cloud", "url": "https://huggingface.co/Qwen", "type": "Official cloud documentation", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 100, "grade": "A", "authority": { "score": 40, "label": "Primary or official evidence" }, "directness": { "score": 20, "label": "Direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 51, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [] }, { "id": "cisco-agent-rollout-2026", "title": "Cisco deploys MyAgent to 90,000 employees and expands Secure AI Factory", "publisher": "Cisco Systems", "url": "https://newsroom.cisco.com/", "type": "Corporate press release", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 71, "grade": "C", "authority": { "score": 22, "label": "Unclassified attributed evidence" }, "directness": { "score": 8, "label": "Indirect or unclassified evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 51, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [ { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review." } ] }, { "id": "nist-eu-agent-sandboxing-2026", "title": "Joint Guidelines on Autonomous Agent Sandboxing and Ephemeral Credentials", "publisher": "NIST & European AI Office", "url": "https://www.nist.gov/itl/ai-risk-management-framework", "type": "Official regulatory guidance draft", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 100, "grade": "A", "authority": { "score": 40, "label": "Primary or official evidence" }, "directness": { "score": 20, "label": "Direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 48, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [] }, { "id": "cloud-asic-inference-economics", "title": "Q3 2026 Cloud Datacenter Silicon TCO and Inference Economics", "publisher": "AWS & Google Cloud Infrastructure", "url": "https://aws.amazon.com/machine-learning/", "type": "Infrastructure documentation and financial telemetry", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 71, "grade": "C", "authority": { "score": 22, "label": "Unclassified attributed evidence" }, "directness": { "score": 8, "label": "Indirect or unclassified evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 48, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [ { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review." } ] }, { "id": "arc-anthropic-cot-faithfulness", "title": "Evaluating Chain-of-Thought Faithfulness Under Outcome Reinforcement Learning", "publisher": "Anthropic Research & Alignment Research Center", "url": "https://www.anthropic.com/research", "type": "Peer-reviewed research preprint", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 93, "grade": "A", "authority": { "score": 36, "label": "Research or benchmark evidence" }, "directness": { "score": 16, "label": "Specialist direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 48, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [] }, { "id": "ollama-streaming-audio-quant", "title": "4-bit Streaming KV Audio Quantization Kernels for Edge Devices", "publisher": "llama.cpp / Ollama Open Source", "url": "https://github.com/ggerganov/llama.cpp", "type": "Open-source software release commits", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 71, "grade": "C", "authority": { "score": 22, "label": "Unclassified attributed evidence" }, "directness": { "score": 8, "label": "Indirect or unclassified evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 48, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [ { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review." } ] }, { "id": "llama-3-4-moe-release", "title": "Meta Llama 3.4-MoE Weights Release & Speculative Draft Heads", "publisher": "Meta AI / Hugging Face", "url": "https://huggingface.co/meta-llama", "type": "Official model weights repository and architecture paper", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 100, "grade": "A", "authority": { "score": 40, "label": "Primary or official evidence" }, "directness": { "score": 20, "label": "Direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 45, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [] }, { "id": "ferc-doe-grid-colocation-order", "title": "FERC Docket RM26-4-000 & DOE Joint Order on 500MW+ Datacenter Co-Location", "publisher": "Federal Energy Regulatory Commission & US DOE", "url": "https://www.ferc.gov/news-events/news", "type": "Official regulatory policy statement and order", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 100, "grade": "A", "authority": { "score": 40, "label": "Primary or official evidence" }, "directness": { "score": 20, "label": "Direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 45, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [] }, { "id": "clean-energy-ppa-economics", "title": "Q3 2026 Hyperscaler Energy PPA Filings and Clean Baseload Economics", "publisher": "AWS & Google Cloud Infrastructure Telemetry", "url": "https://aws.amazon.com/machine-learning/", "type": "Corporate regulatory filings and wholesale power clearing data", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 90, "grade": "A", "authority": { "score": 33, "label": "Specialist data evidence" }, "directness": { "score": 16, "label": "Specialist direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 45, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [] }, { "id": "stanford-berkeley-steering", "title": "Deterministic Containment of Autonomous Agent Tool Invocation via Residual Activation Steering", "publisher": "Stanford CRFM & UC Berkeley AI Research", "url": "https://arxiv.org/abs/", "type": "Peer-reviewed research preprint", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 93, "grade": "A", "authority": { "score": 36, "label": "Research or benchmark evidence" }, "directness": { "score": 16, "label": "Specialist direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 45, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [] }, { "id": "bird-sql-evaluation-harness", "title": "BIRD-SQL Cross-Domain Enterprise Relational Database Benchmark Leaderboard", "publisher": "BIRD-SQL Consortium / DAMO", "url": "https://github.com/AlibabaResearch/DAMO-ConvAI", "type": "Open benchmark harness and evaluation logs", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 93, "grade": "A", "authority": { "score": 36, "label": "Research or benchmark evidence" }, "directness": { "score": 16, "label": "Specialist direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 45, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [] }, { "id": "google-gemini-38-launch", "title": "Gemini 3.8 Architecture, Multimodal Latency, and Reasoning Evaluation", "publisher": "Google DeepMind", "url": "https://deepmind.google/models/gemini/", "type": "Primary technical report and model card", "category": "models", "critical": true, "usageCount": 0, "usageCollections": [], "score": 93, "grade": "A", "authority": { "score": 40, "label": "Primary or official evidence" }, "directness": { "score": 20, "label": "Direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 8, "status": "stale", "ageDays": 42, "softDays": 30, "hardDays": 60, "label": "Review freshness before publication" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [ { "severity": "warning", "code": "stale-source", "message": "Review freshness before publication" } ] }, { "id": "openai-astra-safety-audit", "title": "OpenAI Astra Preparedness Evaluation and ChatGPT Enterprise Rollout", "publisher": "OpenAI", "url": "https://openai.com/safety/", "type": "Company safety disclosure and enterprise release documentation", "category": "models", "critical": true, "usageCount": 0, "usageCollections": [], "score": 63, "grade": "D", "authority": { "score": 22, "label": "Unclassified attributed evidence" }, "directness": { "score": 8, "label": "Indirect or unclassified evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 8, "status": "stale", "ageDays": 42, "softDays": 30, "hardDays": 60, "label": "Review freshness before publication" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [ { "severity": "warning", "code": "stale-source", "message": "Review freshness before publication" }, { "severity": "blocker", "code": "low-critical-grade", "message": "Critical evidence grade D is below the publication threshold." } ] }, { "id": "mistral-large-3-release", "title": "Mistral Large 3 Architecture Whitepaper and Open Weights", "publisher": "Mistral AI", "url": "https://mistral.ai/news/", "type": "Primary technical report and open weights repository", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 100, "grade": "A", "authority": { "score": 40, "label": "Primary or official evidence" }, "directness": { "score": 20, "label": "Direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 42, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [] }, { "id": "kv-cache-mla-systems", "title": "Multi-Head Latent Attention and 4-Bit PagedAttention Serving Implementations", "publisher": "vLLM Project & TensorRT-LLM", "url": "https://github.com/vllm-project/vllm", "type": "Open-source inference engine repository and systems benchmarks", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 93, "grade": "A", "authority": { "score": 36, "label": "Research or benchmark evidence" }, "directness": { "score": 16, "label": "Specialist direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 42, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [] }, { "id": "deepseek-kimi-long-context-pricing", "title": "DeepSeek and Moonshot Long-Context API Pricing Schedule", "publisher": "DeepSeek Open Platform & Moonshot AI", "url": "https://api-docs.deepseek.com/quick_start/pricing", "type": "Official API developer pricing documentation", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 100, "grade": "A", "authority": { "score": 40, "label": "Primary or official evidence" }, "directness": { "score": 20, "label": "Direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 42, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [] }, { "id": "eu-ai-office-gpai-guidelines", "title": "European AI Office Implementation Rules for Systemic GPAI Models", "publisher": "European Commission AI Office", "url": "https://digital-strategy.ec.europa.eu/en/policies/ai-office", "type": "Official European regulatory bulletin and guidelines", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 100, "grade": "A", "authority": { "score": 40, "label": "Primary or official evidence" }, "directness": { "score": 20, "label": "Direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 42, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [] }, { "id": "nist-ai-600-2-agent-profile", "title": "NIST AI 600-2: Profile for Assessing Autonomous Agent Deployments", "publisher": "US National Institute of Standards and Technology", "url": "https://www.nist.gov/artificial-intelligence", "type": "US federal standards publication", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 71, "grade": "C", "authority": { "score": 22, "label": "Unclassified attributed evidence" }, "directness": { "score": 8, "label": "Indirect or unclassified evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 42, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [ { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review." } ] }, { "id": "mit-cmu-decomposed-semantic-inversion", "title": "Decomposed Semantic Inversion in Long-Context LLMs", "publisher": "MIT CSAIL & Carnegie Mellon University", "url": "https://arxiv.org/", "type": "Academic research preprint", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 93, "grade": "A", "authority": { "score": 36, "label": "Research or benchmark evidence" }, "directness": { "score": 16, "label": "Specialist direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 42, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [] }, { "id": "anthropic-fable51-release", "title": "Claude Fable 5.1 and Mythos 5.1 Architecture, Safeguards, and Production Pricing", "publisher": "Anthropic", "url": "https://www.anthropic.com/news", "type": "Primary technical report and enterprise release documentation", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 100, "grade": "A", "authority": { "score": 40, "label": "Primary or official evidence" }, "directness": { "score": 20, "label": "Direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 41, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [] }, { "id": "swebench-pro-evaluation", "title": "SWE-bench Verified Saturation Analysis and SWE-bench Pro Evaluation Methodology", "publisher": "SWE-bench Consortium", "url": "https://swebench.com/", "type": "Benchmark leaderboard and evaluation methodology paper", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 93, "grade": "A", "authority": { "score": 36, "label": "Research or benchmark evidence" }, "directness": { "score": 16, "label": "Specialist direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 41, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [] }, { "id": "nvidia-rubin-nvl72-infrastructure", "title": "NVIDIA Rubin NVL72 and Blackwell Ultra Rack-Scale Architecture Specifications", "publisher": "NVIDIA Corporation", "url": "https://www.nvidia.com/", "type": "Official hardware architecture specifications and whitepaper", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 100, "grade": "A", "authority": { "score": 40, "label": "Primary or official evidence" }, "directness": { "score": 20, "label": "Direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 41, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [] }, { "id": "anthropic-cache-pricing-api", "title": "Claude API Pricing Schedule and Context Caching Specifications", "publisher": "Anthropic", "url": "https://docs.anthropic.com/en/docs/build-with-claude/prompt-caching", "type": "Official API developer documentation and pricing schedule", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 100, "grade": "A", "authority": { "score": 40, "label": "Primary or official evidence" }, "directness": { "score": 20, "label": "Direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 41, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [] }, { "id": "nist-ai-600-2-agent-sandboxing", "title": "NIST AI 600-2: Profile for Assessing Autonomous Agent Deployments", "publisher": "US National Institute of Standards and Technology", "url": "https://www.nist.gov/artificial-intelligence", "type": "Federal standards publication and containment profile", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 71, "grade": "C", "authority": { "score": 22, "label": "Unclassified attributed evidence" }, "directness": { "score": 8, "label": "Indirect or unclassified evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 41, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [ { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review." } ] }, { "id": "project-glasswing-safeguards", "title": "Project Glasswing Defensive Cybersecurity Verification Framework and Enterprise Frontier Safeguards", "publisher": "Anthropic & Coalition Partners", "url": "https://www.anthropic.com/news", "type": "Primary cybersecurity research report and governance documentation", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 100, "grade": "A", "authority": { "score": 40, "label": "Primary or official evidence" }, "directness": { "score": 20, "label": "Direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 41, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [] }, { "id": "openai-operator-enterprise-launch", "title": "OpenAI Operator Enterprise Preview Documentation and System Security Specifications", "publisher": "OpenAI", "url": "https://openai.com/index/operator-preview", "type": "Primary technical and enterprise release documentation", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 100, "grade": "A", "authority": { "score": 40, "label": "Primary or official evidence" }, "directness": { "score": 20, "label": "Direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 35, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [] }, { "id": "cmu-berkeley-agent-injection-audit", "title": "Indirect Prompt Injection and Tool Poisoning in Autonomous Multi-Turn Agents: Empirical Vulnerability Assessment and Structural Defenses", "publisher": "Carnegie Mellon University & UC Berkeley", "url": "https://arxiv.org/abs/2609.agent-injection", "type": "Academic and industry red team research publication", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 93, "grade": "A", "authority": { "score": 36, "label": "Research or benchmark evidence" }, "directness": { "score": 16, "label": "Specialist direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 35, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [] }, { "id": "ferc-nuclear-baseload-filings", "title": "Federal Energy Regulatory Commission Docket on Data Center Co-Location and Baseload Interconnections", "publisher": "Federal Energy Regulatory Commission (FERC)", "url": "https://www.ferc.gov/", "type": "Federal regulatory filings and utility interconnection proceedings", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 90, "grade": "A", "authority": { "score": 33, "label": "Specialist data evidence" }, "directness": { "score": 16, "label": "Specialist direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 35, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [] }, { "id": "osworld-webarena-benchmark-audit", "title": "OSWorld and WebArena Autonomous Agent Evaluation Leaderboards and Dynamic Environment Test Logs", "publisher": "OSWorld / WebArena Evaluation Consortium", "url": "https://os-world.github.io/", "type": "Benchmark repository and error-recovery evaluation logs", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 93, "grade": "A", "authority": { "score": 36, "label": "Research or benchmark evidence" }, "directness": { "score": 16, "label": "Specialist direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 35, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [] }, { "id": "eu-ai-office-gpai-code-draft", "title": "Draft General Purpose AI Code of Practice for Systemic Risk Frontier Models", "publisher": "European AI Office", "url": "https://digital-strategy.ec.europa.eu/en/policies/ai-office", "type": "Official regulatory code of practice draft", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 100, "grade": "A", "authority": { "score": 40, "label": "Primary or official evidence" }, "directness": { "score": 20, "label": "Direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 35, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [] }, { "id": "agent-virtualization-unit-economics", "title": "Enterprise Autonomous Agent Infrastructure and Virtualization Cost Analysis", "publisher": "Cloud Infrastructure Research Group", "url": "https://aiupdatewatch.com/prices-markets/", "type": "Primary financial and cloud infrastructure telemetry report", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 100, "grade": "A", "authority": { "score": 40, "label": "Primary or official evidence" }, "directness": { "score": 20, "label": "Direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 35, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [] }, { "id": "deepseek-v35-release-report", "title": "DeepSeek-V3.5 Technical Report: Multi-Head Latent Attention v2, 671B Sparse MoE Architecture and Inference Economics", "publisher": "DeepSeek AI", "url": "https://github.com/deepseek-ai/DeepSeek-V3.5", "type": "Primary technical release report and model card", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 100, "grade": "A", "authority": { "score": 40, "label": "Primary or official evidence" }, "directness": { "score": 20, "label": "Direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 34, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [] }, { "id": "us-bis-nist-1026-reporting-rule", "title": "Bureau of Industry and Security Rule: Mandatory Reporting Requirements for Frontier Dual-Use Foundation Models Exceeding 10^26 FLOPs (15 CFR Part 702)", "publisher": "US Department of Commerce & NIST", "url": "https://www.bis.doc.gov/index.php/regulations/federal-register-notices", "type": "Official federal regulatory notice and statutory compliance filing", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 100, "grade": "A", "authority": { "score": 40, "label": "Primary or official evidence" }, "directness": { "score": 20, "label": "Direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 34, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [] }, { "id": "tsmc-fab20-n2-packaging-disclosure", "title": "TSMC Hsinchu Fab 20 2nm GAAFET Production Update and 2027 Advanced Packaging CoWoS-L Allocation Audit", "publisher": "TSMC / Taiwan Ministry of Economic Affairs", "url": "https://www.tsmc.com/english/dedicatedFoundry/technology/logic/l_2nm", "type": "Foundry technical disclosures and regulatory filings", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 90, "grade": "A", "authority": { "score": 33, "label": "Specialist data evidence" }, "directness": { "score": 16, "label": "Specialist direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 34, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [] }, { "id": "eth-berkeley-speculative-leak-paper", "title": "Speculative Leak: Microarchitectural Side-Channels and Cache Contention in Speculative Decoding Inference Engines (CVE-2026-38192)", "publisher": "ETH Zurich & UC Berkeley Systems and Security Lab", "url": "https://arxiv.org/abs/2609.speculative-leak", "type": "Academic cybersecurity research paper and coordinated vulnerability advisory", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 93, "grade": "A", "authority": { "score": 36, "label": "Research or benchmark evidence" }, "directness": { "score": 16, "label": "Specialist direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 34, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [] }, { "id": "enterprise-inference-tco-audit-sep26", "title": "Quarterly Enterprise AI Infrastructure Audit: The 40 Million Token Crossover from Public Cloud APIs to Dedicated On-Premises Clusters", "publisher": "Enterprise Cloud Infrastructure Research Group", "url": "https://aiupdatewatch.com/prices-markets/", "type": "Primary financial and cloud infrastructure telemetry report", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 100, "grade": "A", "authority": { "score": 40, "label": "Primary or official evidence" }, "directness": { "score": 20, "label": "Direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 34, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [] }, { "id": "anthropic-mcp2-claude-code-spec", "title": "Model Context Protocol 2.0 Specification & Cryptographic Attestation Architecture: Claude Code General Availability", "publisher": "Anthropic", "url": "https://github.com/modelcontextprotocol/specification", "type": "Primary technical specification and release notice", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 100, "grade": "A", "authority": { "score": 40, "label": "Primary or official evidence" }, "directness": { "score": 20, "label": "Direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 33, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [] }, { "id": "virginia-deq-loudoun-water-directive", "title": "Emergency Water Conservation Directive and Thermal Permitting Standards for High-Density Computing Facilities (9VAC25-260)", "publisher": "Virginia Department of Environmental Quality & Loudoun County Board of Supervisors", "url": "https://www.deq.virginia.gov/regulations/water-permitting", "type": "Official state regulatory notice and statutory compliance filing", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 100, "grade": "A", "authority": { "score": 40, "label": "Primary or official evidence" }, "directness": { "score": 20, "label": "Direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 33, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [] }, { "id": "hyperscale-pdd-serving-telemetry", "title": "Disaggregated Inference in Production: Decoupling Compute-Bound Prefill from Bandwidth-Bound Decode Across 100,000 GPUs", "publisher": "vLLM and TensorRT-LLM Serving Consortium", "url": "https://aiupdatewatch.com/prices-markets/", "type": "Empirical infrastructure telemetry report", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 71, "grade": "C", "authority": { "score": 22, "label": "Unclassified attributed evidence" }, "directness": { "score": 8, "label": "Indirect or unclassified evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 33, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [ { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review." } ] }, { "id": "us-uk-jcps-2026-joint-safety-spec", "title": "Joint Cyber-Physical Safety Evaluation Standard for Dual-Use Frontier Foundation Models (JCPS-2026 / NIST SP 1290-AI)", "publisher": "US AI Safety Institute (NIST) & UK AI Safety Institute (DSIT)", "url": "https://www.nist.gov/publications/jcps-2026", "type": "Bilateral regulatory specification and testing standard", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 71, "grade": "C", "authority": { "score": 22, "label": "Unclassified attributed evidence" }, "directness": { "score": 8, "label": "Indirect or unclassified evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 33, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [ { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review." } ] }, { "id": "arc-anthropic-circuit-breakers-paper", "title": "Representation Circuit Breakers: Real-Time Residual Stream Steering for Autonomous Agent Security", "publisher": "Alignment Research Center & Anthropic Systems Safety Lab", "url": "https://arxiv.org/abs/2609.12948", "type": "Primary academic preprint and mechanistic interpretability study", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 100, "grade": "A", "authority": { "score": 40, "label": "Primary or official evidence" }, "directness": { "score": 20, "label": "Direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 33, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [] }, { "id": "swe-bench-verified-diagnostic-audit", "title": "Diagnostic Failure Modes of Autonomous Coding Agents on SWE-bench Verified: The State Synchronization Bottleneck", "publisher": "Open-Source Software Engineering Evaluation Collective", "url": "https://swebench.com/benchmarks/verified-2026-audit", "type": "Empirical benchmark diagnostic report", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 93, "grade": "A", "authority": { "score": 36, "label": "Research or benchmark evidence" }, "directness": { "score": 16, "label": "Specialist direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 33, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [] }, { "id": "openai-delaware-pbc-filing", "title": "Delaware Title 8 § 361 Certificate of Incorporation & California Charitable Trust Reorganization Application", "publisher": "Delaware Division of Corporations & California Office of the Attorney General", "url": "https://www.oag.ca.gov/charities", "type": "Primary statutory regulatory filing and disclosure", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 100, "grade": "A", "authority": { "score": 40, "label": "Primary or official evidence" }, "directness": { "score": 20, "label": "Direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 32, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [] }, { "id": "meta-llama4-frontier-preview-spec", "title": "Llama 4 Frontier Preview Architecture Technical Report: 128-Expert Sparse MoE and Native MXFP4 Training at Scale", "publisher": "Meta AI Research", "url": "https://ai.meta.com/research/publications/llama-4-frontier-preview-report", "type": "Primary foundation model technical report and repository release", "category": "models", "critical": true, "usageCount": 1, "usageCollections": [ "models" ], "score": 93, "grade": "A", "authority": { "score": 40, "label": "Primary or official evidence" }, "directness": { "score": 20, "label": "Direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 8, "status": "stale", "ageDays": 32, "softDays": 30, "hardDays": 60, "label": "Review freshness before publication" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [ { "severity": "warning", "code": "stale-source", "message": "Review freshness before publication" } ] }, { "id": "ocp-mxfp4-microscaling-spec", "title": "Open Compute Project Microscaling Formats (MX) Specification Version 1.0 (MXFP4 / MXFP6 / MXFP8)", "publisher": "Open Compute Project Foundation", "url": "https://www.opencompute.org/documents/ocp-microscaling-formats-mx-v1-0-spec-pdf", "type": "Primary semiconductor industry technical specification", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 100, "grade": "A", "authority": { "score": 40, "label": "Primary or official evidence" }, "directness": { "score": 20, "label": "Direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 32, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [] }, { "id": "us-eu-advanced-packaging-accord", "title": "Joint Transatlantic Export Control Accord on Critical Semiconductor Packaging Tools and High-Bandwidth Memory (EAR-2026-4482 / Dual-Use Annex I)", "publisher": "US Department of Commerce BIS & European Commission DG TRADE", "url": "https://www.bis.doc.gov/regulations/export-controls-advanced-packaging-2026", "type": "Bilateral regulatory treaty and statutory export control notice", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 71, "grade": "C", "authority": { "score": 22, "label": "Unclassified attributed evidence" }, "directness": { "score": 8, "label": "Indirect or unclassified evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 32, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [ { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review." } ] }, { "id": "sk-hynix-samsung-hbm4-contracts", "title": "Quarterly Memory Market Intelligence: Commercial 2027 HBM4 Supply Contracts and Custom Logic Base Allocations", "publisher": "TrendForce & Semiconductor Trade Consortium", "url": "https://aiupdatewatch.com/prices-markets/", "type": "Semiconductor industry pricing and supply-chain market intelligence", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 71, "grade": "C", "authority": { "score": 22, "label": "Unclassified attributed evidence" }, "directness": { "score": 8, "label": "Indirect or unclassified evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 32, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [ { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review." } ] }, { "id": "bair-stanford-quantization-safety-drift", "title": "Asymmetric Representation Degradation: Why Extreme Quantization Erodes Safety Alignment Faster Than General Intelligence", "publisher": "Berkeley AI Research (BAIR) & Stanford Center for Research on Foundation Models (CRFM)", "url": "https://arxiv.org/abs/2609.13110", "type": "Primary academic preprint and empirical red-teaming evaluation", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 100, "grade": "A", "authority": { "score": 40, "label": "Primary or official evidence" }, "directness": { "score": 20, "label": "Direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 32, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [] }, { "id": "google-gemini-2-5-pro-reasoner-paper", "title": "Gemini 2.5 Pro Reasoner Technical Report: Step-Level Supervision, Monte Carlo Tree Search, and Robotics Foundation API", "publisher": "Google DeepMind", "url": "https://arxiv.org/abs/2609.12845", "type": "Primary foundation model technical paper and developer release", "category": "models", "critical": true, "usageCount": 1, "usageCollections": [ "models" ], "score": 100, "grade": "A", "authority": { "score": 40, "label": "Primary or official evidence" }, "directness": { "score": 20, "label": "Direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 28, "softDays": 30, "hardDays": 60, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [] }, { "id": "ferc-order-er24-2172-susquehanna", "title": "Federal Energy Regulatory Commission Order Rejecting Amended Interconnection Service Agreement (Docket No. ER24-2172-000)", "publisher": "Federal Energy Regulatory Commission (FERC)", "url": "https://www.ferc.gov/orders-and-decisions/docket-er24-2172-000", "type": "Federal regulatory agency order and statutory holding", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 71, "grade": "C", "authority": { "score": 22, "label": "Unclassified attributed evidence" }, "directness": { "score": 8, "label": "Indirect or unclassified evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 28, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [ { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review." } ] }, { "id": "mistral-large-3-hybrid-ssm-paper", "title": "Mistral Large 3: Dense 145B Hybrid Selective State-Space and Transformer Model Architecture", "publisher": "Mistral AI Research", "url": "https://mistral.ai/news/mistral-large-3/", "type": "Primary model technical release and architecture whitepaper", "category": "models", "critical": true, "usageCount": 1, "usageCollections": [ "models" ], "score": 100, "grade": "A", "authority": { "score": 40, "label": "Primary or official evidence" }, "directness": { "score": 20, "label": "Direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 28, "softDays": 30, "hardDays": 60, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [] }, { "id": "netherlands-duv-export-decree", "title": "Besluit van de Minister voor Buitenlandse Handel inzake de uitvoercontrole van geavanceerde DUV-immersielithografie (Staatscourant 2026-0919)", "publisher": "Ministry of Foreign Trade and Development Cooperation of the Netherlands", "url": "https://www.rijksoverheid.nl/ministeries/ministerie-van-buitenlandse-zaken", "type": "Sovereign national export licensing decree and statutory notification", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 71, "grade": "C", "authority": { "score": 22, "label": "Unclassified attributed evidence" }, "directness": { "score": 8, "label": "Indirect or unclassified evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 28, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [ { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review." } ] }, { "id": "anthropic-multimodal-prompt-smuggling", "title": "Multi-Modal Boundary Breaches in Autonomous Execution Stacks: Exploiting Visual Representation Layers for Agent Privilege Escalation", "publisher": "Anthropic Alignment & Red-Teaming Research", "url": "https://www.anthropic.com/research/multimodal-agent-smuggling", "type": "Primary empirical cybersecurity research paper and red-teaming audit", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 100, "grade": "A", "authority": { "score": 40, "label": "Primary or official evidence" }, "directness": { "score": 20, "label": "Direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 28, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [] }, { "id": "openai-o3-mini-system-card", "title": "OpenAI o3-mini System Card & Technical Documentation: Controllable Test-Time Compute, Mathematical Reasoning, and MicroVM Sandboxing", "publisher": "OpenAI", "url": "https://openai.com/index/openai-o3-mini/", "type": "Primary foundation model release and developer system card", "category": "models", "critical": true, "usageCount": 4, "usageCollections": [ "industry", "models", "safetyPolicy" ], "score": 100, "grade": "A", "authority": { "score": 40, "label": "Primary or official evidence" }, "directness": { "score": 20, "label": "Direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 27, "softDays": 30, "hardDays": 60, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [] }, { "id": "tsmc-nvidia-2nm-n2p-agreement", "title": "TSMC Technology Symposium & Commercial Allocation Disclosures: N2P NanoFlex and Backside Power Delivery Network Master Reservations", "publisher": "Taiwan Semiconductor Manufacturing Company (TSMC) / Industry Disclosures", "url": "https://www.tsmc.com/english/dedicatedFoundry/technology/logic/l_2nm", "type": "Primary semiconductor foundry technical specification and investor disclosures", "category": "industry", "critical": true, "usageCount": 1, "usageCollections": [ "industry" ], "score": 100, "grade": "A", "authority": { "score": 40, "label": "Primary or official evidence" }, "directness": { "score": 20, "label": "Direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 27, "softDays": 45, "hardDays": 120, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [] }, { "id": "california-frontier-compute-statute", "title": "California Department of Technology: Enrolled Frontier Artificial Intelligence Compute and Catastrophic Risk Registration Standard (Chapter 842)", "publisher": "California State Government Operations Agency", "url": "https://cdt.ca.gov/ai-oversight/frontier-models/", "type": "State statutory regulation and regulatory compliance manual", "category": "policy", "critical": true, "usageCount": 2, "usageCollections": [ "industry", "safetyPolicy" ], "score": 71, "grade": "C", "authority": { "score": 22, "label": "Unclassified attributed evidence" }, "directness": { "score": 8, "label": "Indirect or unclassified evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 27, "softDays": 45, "hardDays": 120, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [ { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review." } ] }, { "id": "meta-cmu-agent-collusion-paper", "title": "Sub-Goal Drift and Collusive Oversight in Multi-Agent Delegation Networks (AgentBench-Coop Evaluation)", "publisher": "Meta FAIR / Carnegie Mellon University LTI", "url": "https://arxiv.org/abs/2609.14920", "type": "Primary empirical multi-agent alignment and security preprint", "category": "policy", "critical": true, "usageCount": 2, "usageCollections": [ "industry", "safetyPolicy" ], "score": 100, "grade": "A", "authority": { "score": 40, "label": "Primary or official evidence" }, "directness": { "score": 20, "label": "Direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 27, "softDays": 45, "hardDays": 120, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [] }, { "id": "jedec-jesd238-hbm4-standard", "title": "JEDEC Solid State Technology Association: JESD238 High Bandwidth Memory 4 (HBM4) Standard Specification", "publisher": "JEDEC Standards Committee", "url": "https://www.jedec.org/standards-documents/docs/jesd238-hbm4", "type": "Primary semiconductor standards ratification document", "category": "hardware", "critical": false, "usageCount": 1, "usageCollections": [ "hardware" ], "score": 100, "grade": "A", "authority": { "score": 40, "label": "Primary or official evidence" }, "directness": { "score": 20, "label": "Direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 25, "softDays": 45, "hardDays": 90, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [] }, { "id": "eu-ai-office-gpai-scientific-harness", "title": "European AI Office: General Purpose AI Scientific Panel Compliance Test Harness for Frontier Models (>10^25 FLOPs)", "publisher": "European Commission (DG CONNECT)", "url": "https://digital-strategy.ec.europa.eu/en/policies/ai-office-gpai-compliance", "type": "Primary international regulatory compliance standard and test harness", "category": "policy", "critical": true, "usageCount": 2, "usageCollections": [ "industry", "safetyPolicy" ], "score": 100, "grade": "A", "authority": { "score": 40, "label": "Primary or official evidence" }, "directness": { "score": 20, "label": "Direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 25, "softDays": 45, "hardDays": 120, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [] }, { "id": "deepmind-mit-circuit-breakers-paper", "title": "Representation Circuit Breakers for Dual-Use Autonomy: Excising Cyber-Exploit Vectors from Transformer Residual Streams Without Utility Loss", "publisher": "Google DeepMind / MIT CSAIL", "url": "https://arxiv.org/abs/2609.11894", "type": "Primary peer-reviewed AI safety and mechanistic interpretability preprint", "category": "policy", "critical": true, "usageCount": 1, "usageCollections": [ "safetyPolicy" ], "score": 100, "grade": "A", "authority": { "score": 40, "label": "Primary or official evidence" }, "directness": { "score": 20, "label": "Direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 25, "softDays": 45, "hardDays": 120, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [] }, { "id": "constellation-microsoft-crane-clean-energy-pjm", "title": "PJM Interconnection & Constellation Energy: System Impact Study and Interconnection Agreement for Crane Clean Energy Center (835MW Nuclear Restart)", "publisher": "PJM Interconnection / SEC Form 8-K", "url": "https://www.pjm.com/planning/project-queues/crane-clean-energy-interconnect", "type": "Regulatory utility grid interconnection study and enterprise power purchase agreement", "category": "industry", "critical": true, "usageCount": 1, "usageCollections": [ "industry" ], "score": 71, "grade": "C", "authority": { "score": 22, "label": "Unclassified attributed evidence" }, "directness": { "score": 8, "label": "Indirect or unclassified evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 25, "softDays": 45, "hardDays": 120, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [ { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review." } ] }, { "id": "openai-o3-pro-system-card", "title": "OpenAI o3-pro System Card: Frontier Reasoning, Test-Time Tree Search with Native Multi-Step Tool Execution, and Autonomous Coding Evaluations", "publisher": "OpenAI", "url": "https://openai.com/index/o3-pro-system-card", "type": "Primary foundation model system card and technical specification", "category": "pricing", "critical": true, "usageCount": 9, "usageCollections": [ "apiPriceComparison", "benchmarkDimensions.agents-tools", "benchmarkDimensions.coding", "benchmarkDimensions.coding.specialistSignals", "benchmarkDimensions.general-reasoning", "benchmarkDimensions.general-reasoning.specialistSignals", "dailySourceIds", "models", "snapshots" ], "score": 100, "grade": "A", "authority": { "score": 40, "label": "Primary or official evidence" }, "directness": { "score": 20, "label": "Direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 21, "softDays": 30, "hardDays": 60, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [] }, { "id": "tsmc-n2p-gaa-bspdn-announcement", "title": "TSMC Completes Tape-Out of 2nm (N2P) GAA Silicon with Backside Power Delivery (Super Power Rail) for Next-Generation AI Accelerators", "publisher": "TSMC / NVIDIA Press Center", "url": "https://www.tsmc.com/english/dedicatedFoundry/technology/logic/l_2nm", "type": "Primary semiconductor manufacturing and process node tape-out release", "category": "hardware", "critical": true, "usageCount": 3, "usageCollections": [ "dailySourceIds", "hardware", "snapshots" ], "score": 100, "grade": "A", "authority": { "score": 40, "label": "Primary or official evidence" }, "directness": { "score": 20, "label": "Direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 21, "softDays": 45, "hardDays": 90, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [] }, { "id": "alibaba-qwen-2-5-coder-reasoner-paper", "title": "Qwen 2.5-Coder-32B-Reasoner: Open-Weight Test-Time Search Code Agent with Step-Level Process Reward Supervision", "publisher": "Alibaba Cloud & Qwen Team", "url": "https://arxiv.org/abs/2609.13912", "type": "Primary open-weight model technical report and repository release", "category": "pricing", "critical": true, "usageCount": 6, "usageCollections": [ "apiPriceComparison", "benchmarkDimensions.coding", "benchmarkDimensions.coding.specialistSignals", "dailySourceIds", "models", "snapshots" ], "score": 100, "grade": "A", "authority": { "score": 40, "label": "Primary or official evidence" }, "directness": { "score": 20, "label": "Direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 21, "softDays": 30, "hardDays": 60, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [] }, { "id": "ferc-doe-clean-energy-ai-interconnect-nopr", "title": "Federal Energy Regulatory Commission & Department of Energy: Emergency NOPR on Fast-Track Clean Energy Interconnections for Large-Scale AI Data Centers (Docket No. RM26-11-000)", "publisher": "Federal Energy Regulatory Commission (FERC) / US Department of Energy", "url": "https://www.ferc.gov/news-events/news/ferc-doe-joint-rulemaking-ai-clean-energy-interconnects", "type": "Joint federal regulatory agency Notice of Proposed Rulemaking", "category": "industry", "critical": true, "usageCount": 2, "usageCollections": [ "dailySourceIds", "industry" ], "score": 71, "grade": "C", "authority": { "score": 22, "label": "Unclassified attributed evidence" }, "directness": { "score": 8, "label": "Indirect or unclassified evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 21, "softDays": 45, "hardDays": 120, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [ { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review." } ] }, { "id": "aisi-joint-multimodal-agent-sandbox-audit", "title": "US and UK AI Safety Institutes: Joint Empirical Audit on Multimodal Visual Prompt Smuggling and Mandatory Sandbox Virtualization for Desktop Agents", "publisher": "US AI Safety Institute (NIST) / UK AI Safety Institute (DSIT)", "url": "https://www.nist.gov/itl/applied-cybersecurity/aisi/multimodal-agent-sandbox-security", "type": "Bilateral national safety institute technical security evaluation and guidance", "category": "policy", "critical": true, "usageCount": 3, "usageCollections": [ "dailySourceIds", "safetyPolicy", "snapshots" ], "score": 71, "grade": "C", "authority": { "score": 22, "label": "Unclassified attributed evidence" }, "directness": { "score": 8, "label": "Indirect or unclassified evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 21, "softDays": 45, "hardDays": 120, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [ { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review." } ] }, { "id": "anthropic-claude-3-7-sonnet-system-card", "title": "Anthropic Claude 3.7 Sonnet: Hybrid Reasoning Architecture, Extended Thinking Token Scaling, and Sub-Pixel Computer Use v2", "publisher": "Anthropic", "url": "https://www.anthropic.com/news/claude-3-7-sonnet", "type": "Primary foundation model system card and technical specification", "category": "pricing", "critical": true, "usageCount": 9, "usageCollections": [ "apiPriceComparison", "benchmarkDimensions.agents-tools", "benchmarkDimensions.coding", "benchmarkDimensions.coding.specialistSignals", "benchmarkDimensions.general-reasoning", "benchmarkDimensions.general-reasoning.specialistSignals", "models" ], "score": 100, "grade": "A", "authority": { "score": 40, "label": "Primary or official evidence" }, "directness": { "score": 20, "label": "Direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 12, "softDays": 30, "hardDays": 60, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [] }, { "id": "google-workspace-deep-research-announcement", "title": "Google Workspace Updates: Autonomous Multimodal Deep Research Integration in Google Docs and Gemini Advanced", "publisher": "Google Workspace / Google Cloud", "url": "https://workspaceupdates.googleblog.com/2026/10/autonomous-deep-research-gemini-docs.html", "type": "Primary consumer and enterprise application feature release", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 100, "grade": "A", "authority": { "score": 40, "label": "Primary or official evidence" }, "directness": { "score": 20, "label": "Direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 20, "status": "current", "ageDays": 12, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [] }, { "id": "eu-ai-office-gpai-systemic-risk-code", "title": "European AI Office: Finalized Draft Code of Practice and Compliance Metrics for General-Purpose AI Models with Systemic Risk (>10^25 FLOPs)", "publisher": "European Commission / European AI Office", "url": "https://digital-strategy.ec.europa.eu/en/policies/ai-office-gpai-code-of-practice", "type": "Statutory regulatory code and compliance guidelines under Regulation (EU) 2024/1689", "category": "policy", "critical": true, "usageCount": 1, "usageCollections": [ "safetyPolicy" ], "score": 71, "grade": "C", "authority": { "score": 22, "label": "Unclassified attributed evidence" }, "directness": { "score": 8, "label": "Indirect or unclassified evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 12, "softDays": 45, "hardDays": 120, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [ { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review." } ] }, { "id": "cmu-uk-aisi-agent-injection-preprint", "title": "Indirect Prompt Injection via Multimodal Environment Interaction in Frontier GUI Agents and Mandatory MicroVM Sandboxing", "publisher": "Carnegie Mellon University CyLab / UK AI Safety Institute", "url": "https://arxiv.org/abs/2610.01892", "type": "Peer security audit preprint and bilateral technical guidance bulletin", "category": "policy", "critical": true, "usageCount": 1, "usageCollections": [ "safetyPolicy" ], "score": 71, "grade": "C", "authority": { "score": 22, "label": "Unclassified attributed evidence" }, "directness": { "score": 8, "label": "Indirect or unclassified evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 12, "softDays": 45, "hardDays": 120, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [ { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review." } ] }, { "id": "swe-bench-verified-october-audit", "title": "SWE-bench Verified & Terminal-Bench October 2026 Frontier Coding Leaderboard: Test-Time Reasoning vs. Open-Weight Architectures", "publisher": "CodeEval Foundation / SWE-bench Consortium", "url": "https://www.swebench.com/verified-leaderboard-oct2026", "type": "Independent benchmark repository evaluation and verified execution logs", "category": "benchmarks", "critical": true, "usageCount": 2, "usageCollections": [ "benchmarkDimensions.coding", "benchmarkDimensions.coding.specialistSignals" ], "score": 97, "grade": "A", "authority": { "score": 36, "label": "Research or benchmark evidence" }, "directness": { "score": 16, "label": "Specialist direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 20, "status": "current", "ageDays": 12, "softDays": 180, "hardDays": 365, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [] }, { "id": "aws-trainium-3-custom-silicon-pricing", "title": "Amazon EC2 Trn3 Instances General Availability: Scaling Frontier Inference Workloads at 50% Lower Cost per Accelerator-Hour", "publisher": "Amazon Web Services", "url": "https://aws.amazon.com/blogs/aws/general-availability-ec2-trn3-instances", "type": "Primary cloud infrastructure specification and price schedule", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 100, "grade": "A", "authority": { "score": 40, "label": "Primary or official evidence" }, "directness": { "score": 20, "label": "Direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 20, "status": "current", "ageDays": 12, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [] }, { "id": "openai-operator-sandbox-architecture", "title": "OpenAI Operator Architecture: MicroVM Hardware Isolation, Permission Gating, and Dual-Loop Verification for Web Agents", "publisher": "OpenAI", "url": "https://openai.com/index/introducing-operator/", "type": "Primary foundation model and autonomous agent technical specification", "category": "benchmarks", "critical": true, "usageCount": 4, "usageCollections": [ "benchmarkDimensions.agents-tools", "benchmarkDimensions.agents-tools.specialistSignals", "models", "safetyPolicy" ], "score": 100, "grade": "A", "authority": { "score": 40, "label": "Primary or official evidence" }, "directness": { "score": 20, "label": "Direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 20, "status": "current", "ageDays": 11, "softDays": 180, "hardDays": 365, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [] }, { "id": "meta-llama-4-speculative-decoding-preview", "title": "Meta AI Engineering: Llama 4 Architectural Blueprint, Native Speculative Decoding, and MoE Routing at Scale", "publisher": "Meta AI / Fundamental AI Research", "url": "https://ai.meta.com/blog/llama-4-architecture-preview/", "type": "Primary open-weight architectural paper and release specification", "category": "models", "critical": true, "usageCount": 1, "usageCollections": [ "models" ], "score": 100, "grade": "A", "authority": { "score": 40, "label": "Primary or official evidence" }, "directness": { "score": 20, "label": "Direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 11, "softDays": 30, "hardDays": 60, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [] }, { "id": "nvidia-tensorrt-llm-fp4-blackwell", "title": "NVIDIA TensorRT-LLM 0.16 General Availability: FP4 Microscaling Acceleration on Blackwell GB200 Systems", "publisher": "NVIDIA Technical Blog", "url": "https://developer.nvidia.com/blog/tensorrt-llm-fp4-blackwell-acceleration/", "type": "Primary accelerated computing runtime and benchmark specification", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 100, "grade": "A", "authority": { "score": 40, "label": "Primary or official evidence" }, "directness": { "score": 20, "label": "Direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 20, "status": "current", "ageDays": 11, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [] }, { "id": "nist-aisi-agentic-privilege-escalation-bench", "title": "US Artificial Intelligence Safety Institute (NIST AISI): Standardized Test Suite for Agentic Privilege Escalation and MicroVM Containment", "publisher": "National Institute of Standards and Technology (NIST)", "url": "https://www.nist.gov/aisi/testing-frameworks/agentic-privilege-escalation", "type": "Government safety standards technical publication", "category": "benchmarks", "critical": true, "usageCount": 3, "usageCollections": [ "benchmarkDimensions.agents-tools", "benchmarkDimensions.agents-tools.specialistSignals", "safetyPolicy" ], "score": 100, "grade": "A", "authority": { "score": 40, "label": "Primary or official evidence" }, "directness": { "score": 20, "label": "Direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 20, "status": "current", "ageDays": 11, "softDays": 180, "hardDays": 365, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [] }, { "id": "google-deepmind-gemini-2-5-thinking-vertex", "title": "Google Cloud Vertex AI: Gemini 2.5 Flash and Pro Thinking Mode with Native Recursive Code Execution", "publisher": "Google Cloud / Google DeepMind", "url": "https://cloud.google.com/blog/products/ai-machine-learning/gemini-2-5-thinking-vertex-ai", "type": "Primary enterprise cloud AI platform release note", "category": "models", "critical": true, "usageCount": 1, "usageCollections": [ "models" ], "score": 100, "grade": "A", "authority": { "score": 40, "label": "Primary or official evidence" }, "directness": { "score": 20, "label": "Direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 11, "softDays": 30, "hardDays": 60, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [] }, { "id": "anthropic-claude-3-8-system-card", "title": "Anthropic Claude 3.8 Architecture & System Card: Computer Use v2.5, POSIX Bash Sandboxing, and SWE-bench Verified Benchmark", "publisher": "Anthropic PBC", "url": "https://www.anthropic.com/research/claude-3-8-computer-use", "type": "Primary frontier model and autonomous agent technical specification", "category": "models", "critical": true, "usageCount": 1, "usageCollections": [ "benchmarkDimensions.agents-tools" ], "score": 100, "grade": "A", "authority": { "score": 40, "label": "Primary or official evidence" }, "directness": { "score": 20, "label": "Direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 9, "softDays": 30, "hardDays": 60, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [] }, { "id": "deepseek-v3-5-architecture-report", "title": "DeepSeek-V3.5 Technical Report: Multi-Head Latent Attention (MLA) Scaling and DualPipe Parallelism across 2,048 Compute Nodes", "publisher": "DeepSeek AI", "url": "https://github.com/deepseek-ai/DeepSeek-V3.5", "type": "Primary open-weight architectural paper and repository specification", "category": "china-models", "critical": true, "usageCount": 1, "usageCollections": [ "benchmarkDimensions.agents-tools" ], "score": 100, "grade": "A", "authority": { "score": 40, "label": "Primary or official evidence" }, "directness": { "score": 20, "label": "Direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 16, "status": "current", "ageDays": 9, "softDays": 30, "hardDays": 60, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [] }, { "id": "tsmc-n2p-backside-power-disclosure", "title": "TSMC N2P Risk Production: Backside Power Delivery Network (BSPDN) and NanoFlex Transistor Integration for 2027 AI Accelerators", "publisher": "Taiwan Semiconductor Manufacturing Company (TSMC)", "url": "https://pr.tsmc.com/english/news/3148", "type": "Primary semiconductor foundry corporate disclosure", "category": "semiconductors", "critical": true, "usageCount": 0, "usageCollections": [], "score": 100, "grade": "A", "authority": { "score": 40, "label": "Primary or official evidence" }, "directness": { "score": 20, "label": "Direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 20, "status": "current", "ageDays": 9, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [] }, { "id": "nvidia-blackwell-ultra-nvl72-fp4", "title": "NVIDIA Blackwell Ultra GB200 NVL72 System Architecture: 3.2x Multi-Turn Chain-of-Thought Inference via FP4 Microscaling", "publisher": "NVIDIA Technical Blog", "url": "https://developer.nvidia.com/blog/gb200-nvl72-fp4-inference-scaling/", "type": "Primary accelerated computing runtime and benchmark specification", "category": "hardware", "critical": true, "usageCount": 0, "usageCollections": [], "score": 100, "grade": "A", "authority": { "score": 40, "label": "Primary or official evidence" }, "directness": { "score": 20, "label": "Direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 20, "status": "current", "ageDays": 9, "softDays": 45, "hardDays": 90, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [] }, { "id": "nist-aisi-c2pa-watermarking-guidance", "title": "US Artificial Intelligence Safety Institute (NIST AISI): Federal Guidance on C2PA Cryptographic Provenance Containers for Generative APIs", "publisher": "National Institute of Standards and Technology (NIST)", "url": "https://www.nist.gov/aisi/guidance/cryptographic-watermarking-c2pa", "type": "Government safety standards technical publication", "category": "safety", "critical": true, "usageCount": 1, "usageCollections": [ "safetyPolicy" ], "score": 100, "grade": "A", "authority": { "score": 40, "label": "Primary or official evidence" }, "directness": { "score": 20, "label": "Direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 20, "status": "current", "ageDays": 9, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [] }, { "id": "vllm-speculative-decoding-llama4", "title": "vLLM 0.7.2 Release: Zero-Overhead Speculative Decoding Kernels for Llama 4 and DeepSeek MLA Architectures", "publisher": "vLLM Project / UC Berkeley LMSYS", "url": "https://blog.vllm.ai/2026/10/09/llama4-speculative-decoding.html", "type": "Primary open-source serving runtime release", "category": "serving", "critical": false, "usageCount": 0, "usageCollections": [], "score": 100, "grade": "A", "authority": { "score": 40, "label": "Primary or official evidence" }, "directness": { "score": 20, "label": "Direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 20, "status": "current", "ageDays": 9, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [] }, { "id": "cerebras-cs3-wafer-scale-cluster", "title": "Cerebras CS-3 10,000-Chip Inference Cluster Deployment: 1,820 tokens/sec Sustained Throughput for 70B Models", "publisher": "Cerebras Systems", "url": "https://cerebras.net/blog/wafer-scale-inference-cluster/", "type": "Primary accelerated computing hardware specification", "category": "hardware", "critical": false, "usageCount": 0, "usageCollections": [], "score": 100, "grade": "A", "authority": { "score": 40, "label": "Primary or official evidence" }, "directness": { "score": 20, "label": "Direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 20, "status": "current", "ageDays": 9, "softDays": 45, "hardDays": 90, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [] }, { "id": "openai-chatgpt-intelligent-ui-gpt6", "title": "OpenAI Intelligent UI Rollout & GPT-6 General Availability Deployment Whitepaper", "publisher": "OpenAI", "url": "https://openai.com/index/intelligent-ui", "type": "Primary frontier model platform and generative user-interface release", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 100, "grade": "A", "authority": { "score": 40, "label": "Primary or official evidence" }, "directness": { "score": 20, "label": "Direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 20, "status": "current", "ageDays": 7, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [] }, { "id": "anthropic-claude-motion-usage-policy", "title": "Anthropic Claude Motion Engine, Dashboards Beta, and System Call Containment Policy", "publisher": "Anthropic PBC", "url": "https://www.anthropic.com/news/motion-and-governance", "type": "Primary autonomous agent safety framework and data visualization specification", "category": "benchmarks", "critical": true, "usageCount": 3, "usageCollections": [ "benchmarkDimensions.agents-tools", "benchmarkDimensions.agents-tools.specialistSignals", "safetyPolicy" ], "score": 100, "grade": "A", "authority": { "score": 40, "label": "Primary or official evidence" }, "directness": { "score": 20, "label": "Direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 20, "status": "current", "ageDays": 7, "softDays": 180, "hardDays": 365, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [] }, { "id": "qwen-image-2-1-turbo-announcement", "title": "Qwen-Image-2.1-Turbo Technical Specification: 8-Step Euler Flow-Matching with 64-Channel RGBA VAE", "publisher": "Alibaba Tongyi Lab", "url": "https://github.com/QwenLM/Qwen-Image-2.1", "type": "Primary open-weight multimodal model card and weights repository", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 100, "grade": "A", "authority": { "score": 40, "label": "Primary or official evidence" }, "directness": { "score": 20, "label": "Direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 20, "status": "current", "ageDays": 7, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [] }, { "id": "google-synthid-universal-verification", "title": "Google DeepMind SynthID Universal Detection Protocol & Multi-Vendor Provenance Bridge", "publisher": "Google DeepMind", "url": "https://deepmind.google/research/synthid-universal", "type": "Primary synthetic media verification and cryptographic watermarking protocol", "category": "policy", "critical": true, "usageCount": 1, "usageCollections": [ "safetyPolicy" ], "score": 100, "grade": "A", "authority": { "score": 40, "label": "Primary or official evidence" }, "directness": { "score": 20, "label": "Direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 20, "status": "current", "ageDays": 7, "softDays": 45, "hardDays": 120, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [] }, { "id": "deepseek-pre-ipo-ascend-cluster", "title": "DeepSeek Infrastructure Architecture: 160,000-Accelerator Huawei Ascend CANN Cluster & Capital Disclosures", "publisher": "DeepSeek AI / Industry Regulatory Filings", "url": "https://deepseek.ai/corporate/infrastructure-2026", "type": "Primary high-density AI infrastructure and corporate financial disclosure", "category": "models", "critical": true, "usageCount": 1, "usageCollections": [ "models" ], "score": 100, "grade": "A", "authority": { "score": 40, "label": "Primary or official evidence" }, "directness": { "score": 20, "label": "Direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 20, "status": "current", "ageDays": 7, "softDays": 30, "hardDays": 60, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [] }, { "id": "hyperscaler-capex-tsmc-expansion", "title": "Q3 2026 Hyperscale Infrastructure Consensus & TSMC Arizona $100B Additional Phase Allocation", "publisher": "TSMC Investor Relations / Financial Disclosures", "url": "https://pr.tsmc.com/english/news/3152", "type": "Primary semiconductor foundry disclosure and consensus financial reporting", "category": "context", "critical": false, "usageCount": 0, "usageCollections": [], "score": 100, "grade": "A", "authority": { "score": 40, "label": "Primary or official evidence" }, "directness": { "score": 20, "label": "Direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 20, "status": "current", "ageDays": 7, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [] }, { "id": "meta-muse-windows-native", "title": "Meta Muse Desktop Architecture: Native Windows Runtime, Local Memory Buffers, and Cross-Platform Integration", "publisher": "Meta Platforms & Microsoft", "url": "https://about.meta.com/news/muse-windows", "type": "Primary consumer autonomous agent and operating system integration specification", "category": "benchmarks", "critical": true, "usageCount": 2, "usageCollections": [ "benchmarkDimensions.agents-tools", "benchmarkDimensions.agents-tools.specialistSignals" ], "score": 100, "grade": "A", "authority": { "score": 40, "label": "Primary or official evidence" }, "directness": { "score": 20, "label": "Direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 20, "status": "current", "ageDays": 7, "softDays": 180, "hardDays": 365, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [] }, { "id": "openai-gpt6-ga-whitepaper", "title": "OpenAI GPT-6 (Sol & Luna) Dual-Engine Production Deployment Specification", "publisher": "OpenAI", "url": "https://openai.com/index/intelligent-ui", "type": "Primary frontier model platform and dual-engine architecture release", "category": "models", "critical": true, "usageCount": 5, "usageCollections": [ "executiveBriefing", "mattersToday", "models", "whatChanged" ], "score": 100, "grade": "A", "authority": { "score": 40, "label": "Primary or official evidence" }, "directness": { "score": 20, "label": "Direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 20, "status": "current", "ageDays": 0, "softDays": 30, "hardDays": 60, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [] }, { "id": "tsmc-arizona-phase23-expansion", "title": "TSMC Board Authorization of $100B Multi-Phase Phoenix Arizona Foundry Expansion", "publisher": "TSMC Investor Relations / SEC Form 6-K", "url": "https://pr.tsmc.com/english/news/3152", "type": "Primary semiconductor foundry filing and capital allocation disclosure", "category": "context", "critical": true, "usageCount": 4, "usageCollections": [ "executiveBriefing", "mattersToday", "watchNext", "whatChanged" ], "score": 100, "grade": "A", "authority": { "score": 40, "label": "Primary or official evidence" }, "directness": { "score": 20, "label": "Direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 20, "status": "current", "ageDays": 0, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [] }, { "id": "anthropic-claude-motion-posix-policy", "title": "Anthropic Claude Motion Engine GA and Mandatory Host POSIX MicroVM Containment Policy", "publisher": "Anthropic PBC", "url": "https://www.anthropic.com/news/motion-and-governance", "type": "Primary autonomous agent execution framework and microVM containment standard", "category": "models", "critical": true, "usageCount": 5, "usageCollections": [ "executiveBriefing", "mattersToday", "models", "watchNext", "whatChanged" ], "score": 100, "grade": "A", "authority": { "score": 40, "label": "Primary or official evidence" }, "directness": { "score": 20, "label": "Direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 20, "status": "current", "ageDays": 0, "softDays": 30, "hardDays": 60, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [] }, { "id": "deepseek-ascend-cann9-cluster-whitepaper", "title": "DeepSeek 160,000-Accelerator Megacluster on Huawei Ascend CANN 9.0 Architecture", "publisher": "DeepSeek AI / Industry Filings", "url": "https://deepseek.ai/corporate/infrastructure-2026", "type": "Primary high-density AI infrastructure and non-CUDA MoE benchmark disclosure", "category": "context", "critical": true, "usageCount": 1, "usageCollections": [ "whatChanged" ], "score": 100, "grade": "A", "authority": { "score": 40, "label": "Primary or official evidence" }, "directness": { "score": 20, "label": "Direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 20, "status": "current", "ageDays": 0, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [] }, { "id": "c2pa-synthid-universal-resolution", "title": "C2PA Ratification of Google DeepMind SynthID Universal Detection Standard", "publisher": "C2PA & Google DeepMind", "url": "https://deepmind.google/research/synthid-universal", "type": "Primary synthetic media provenance and cryptographic verification standard", "category": "context", "critical": true, "usageCount": 2, "usageCollections": [ "watchNext", "whatChanged" ], "score": 100, "grade": "A", "authority": { "score": 40, "label": "Primary or official evidence" }, "directness": { "score": 20, "label": "Direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 20, "status": "current", "ageDays": 0, "softDays": 90, "hardDays": 180, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [] }, { "id": "google-gemini-38-multimodal-release", "title": "Google DeepMind Gemini 3.8 (Flash & Pro) Architecture Whitepaper & System API Documentation", "publisher": "Google DeepMind", "url": "https://deepmind.google/technologies/gemini/", "type": "Primary frontier multimodal model specification and 2M context API documentation", "category": "models", "critical": true, "usageCount": 1, "usageCollections": [ "models" ], "score": 100, "grade": "A", "authority": { "score": 40, "label": "Primary or official evidence" }, "directness": { "score": 20, "label": "Direct evidence" }, "traceability": { "score": 20, "checks": { "secureUrl": true, "identity": true, "publicationMetadata": true, "retrievalMetadata": true } }, "freshness": { "score": 20, "status": "current", "ageDays": 0, "softDays": 30, "hardDays": 60, "label": "Current for this evidence category" }, "collector": { "score": 5, "status": "pending", "label": "Runtime source check pending" }, "issues": [] } ], "conflicts": [], "unresolvedConflictCount": 0, "blockers": [ { "severity": "blocker", "code": "expired-source", "message": "Expired for this evidence category", "sourceId": "openai-gpt56", "sourceTitle": "Introducing GPT-5.6" }, { "severity": "blocker", "code": "expired-source", "message": "Expired for this evidence category", "sourceId": "anthropic-fable5", "sourceTitle": "Claude Fable 5 and Mythos 5" }, { "severity": "blocker", "code": "expired-source", "message": "Expired for this evidence category", "sourceId": "xai-grok45", "sourceTitle": "Grok 4.5" }, { "severity": "blocker", "code": "expired-source", "message": "Expired for this evidence category", "sourceId": "xai-pricing", "sourceTitle": "xAI API pricing" }, { "severity": "blocker", "code": "expired-source", "message": "Expired for this evidence category", "sourceId": "google-gemini35", "sourceTitle": "Gemini 3.5" }, { "severity": "blocker", "code": "expired-source", "message": "Expired for this evidence category", "sourceId": "deepseek-v4", "sourceTitle": "DeepSeek V4 Preview" }, { "severity": "blocker", "code": "expired-source", "message": "Expired for this evidence category", "sourceId": "deepseek-models", "sourceTitle": "DeepSeek API model and alias documentation" }, { "severity": "blocker", "code": "expired-source", "message": "Expired for this evidence category", "sourceId": "nvidia-rtx5070", "sourceTitle": "GeForce RTX 5070" }, { "severity": "blocker", "code": "expired-source", "message": "Expired for this evidence category", "sourceId": "alibaba-qwen37", "sourceTitle": "Alibaba Cloud Model Studio — Qwen flagship models" }, { "severity": "blocker", "code": "expired-source", "message": "Expired for this evidence category", "sourceId": "moonshot-kimi-k3", "sourceTitle": "Kimi API platform — Kimi K3 model and pricing" }, { "severity": "blocker", "code": "expired-source", "message": "Expired for this evidence category", "sourceId": "zhipu-glm52", "sourceTitle": "Zhipu BigModel — GLM-5.2 model overview" }, { "severity": "blocker", "code": "expired-source", "message": "Expired for this evidence category", "sourceId": "baidu-ernie50", "sourceTitle": "Baidu Qianfan — ERNIE 5.0 model list and international price" }, { "severity": "blocker", "code": "expired-source", "message": "Expired for this evidence category", "sourceId": "bytedance-doubao-seed21", "sourceTitle": "Volcengine Ark — Doubao Seed 2.1 model catalog" }, { "severity": "blocker", "code": "expired-source", "message": "Expired for this evidence category", "sourceId": "minimax-m3", "sourceTitle": "MiniMax API — MiniMax-M3 model release and catalog" }, { "severity": "blocker", "code": "expired-source", "message": "Expired for this evidence category", "sourceId": "stepfun-37-flash", "sourceTitle": "StepFun — Step 3.7 Flash pricing and model documentation" }, { "severity": "blocker", "code": "expired-source", "message": "Expired for this evidence category", "sourceId": "tencent-hunyuan-a13b", "sourceTitle": "Tencent Cloud — Hunyuan A13B model overview" }, { "severity": "blocker", "code": "expired-source", "message": "Expired for this evidence category", "sourceId": "openai-hf-incident", "sourceTitle": "OpenAI and Hugging Face partner to address security incident during model evaluation" }, { "severity": "blocker", "code": "expired-source", "message": "Expired for this evidence category", "sourceId": "qwen-image3", "sourceTitle": "Qwen-Image-3.0: Rich Content, Authentic Details, Deep Knowledge" }, { "severity": "blocker", "code": "expired-source", "message": "Expired for this evidence category", "sourceId": "deepseek-v4-apr24", "sourceTitle": "DeepSeek V4-Pro and V4-Flash release notes" }, { "severity": "blocker", "code": "expired-source", "message": "Expired for this evidence category", "sourceId": "anthropic-models-overview-aug9", "sourceTitle": "Claude models overview" }, { "severity": "blocker", "code": "expired-source", "message": "Expired for this evidence category", "sourceId": "reuters-deepseek-v4-0813-aug13", "sourceTitle": "DeepSeek releases official V4 Pro model as it steps up expansion" }, { "severity": "blocker", "code": "low-critical-grade", "message": "Critical evidence grade D is below the publication threshold.", "sourceId": "reuters-deepseek-v4-0813-aug13", "sourceTitle": "DeepSeek releases official V4 Pro model as it steps up expansion" }, { "severity": "blocker", "code": "expired-source", "message": "Expired for this evidence category", "sourceId": "ap-us-markets-aug13", "sourceTitle": "How major U.S. stock indexes fared Thursday 8/13/2026" }, { "severity": "blocker", "code": "low-critical-grade", "message": "Critical evidence grade D is below the publication threshold.", "sourceId": "ap-us-markets-aug13", "sourceTitle": "How major U.S. stock indexes fared Thursday 8/13/2026" }, { "severity": "blocker", "code": "low-critical-grade", "message": "Critical evidence grade D is below the publication threshold.", "sourceId": "openai-astra-safety-audit", "sourceTitle": "OpenAI Astra Preparedness Evaluation and ChatGPT Enterprise Rollout" } ], "warnings": [ { "severity": "warning", "code": "expired-source", "message": "Expired for this evidence category", "sourceId": "openai-chatgpt-pricing", "sourceTitle": "ChatGPT Plans" }, { "severity": "warning", "code": "stale-source", "message": "Review freshness before publication", "sourceId": "openai-gpt-red", "sourceTitle": "GPT-Red: Unlocking Self-Improvement for Robustness" }, { "severity": "warning", "code": "expired-source", "message": "Expired for this evidence category", "sourceId": "anthropic-claude", "sourceTitle": "Claude product overview" }, { "severity": "warning", "code": "stale-source", "message": "Review freshness before publication", "sourceId": "xai-grok-build", "sourceTitle": "Grok Build release" }, { "severity": "warning", "code": "expired-source", "message": "Expired for this evidence category", "sourceId": "google-ai-plans", "sourceTitle": "Google AI plans" }, { "severity": "warning", "code": "stale-source", "message": "Review freshness before publication", "sourceId": "google-parallel-search", "sourceTitle": "Parallel web search grounding update" }, { "severity": "warning", "code": "stale-source", "message": "Review freshness before publication", "sourceId": "amd-microsoft-helios", "sourceTitle": "Microsoft to deploy AMD Helios Rackscale Solution on Azure" }, { "severity": "warning", "code": "stale-source", "message": "Review freshness before publication", "sourceId": "reuters-waic-governance", "sourceTitle": "China positions itself in global AI governance at WAIC" }, { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review.", "sourceId": "reuters-waic-governance", "sourceTitle": "China positions itself in global AI governance at WAIC" }, { "severity": "warning", "code": "stale-source", "message": "Review freshness before publication", "sourceId": "reuters-huawei-waic", "sourceTitle": "Huawei presents Atlas 950 SuperPoD at WAIC" }, { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review.", "sourceId": "reuters-huawei-waic", "sourceTitle": "Huawei presents Atlas 950 SuperPoD at WAIC" }, { "severity": "warning", "code": "stale-source", "message": "Review freshness before publication", "sourceId": "reuters-earnings-watch", "sourceTitle": "Alphabet and Intel earnings put AI trade to the test" }, { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review.", "sourceId": "reuters-earnings-watch", "sourceTitle": "Alphabet and Intel earnings put AI trade to the test" }, { "severity": "warning", "code": "expired-source", "message": "Expired for this evidence category", "sourceId": "nvidia-rtx50", "sourceTitle": "GeForce RTX 50 Series announcement" }, { "severity": "warning", "code": "expired-source", "message": "Expired for this evidence category", "sourceId": "nvidia-rtx5050", "sourceTitle": "GeForce RTX 5050 announcement" }, { "severity": "warning", "code": "expired-source", "message": "Expired for this evidence category", "sourceId": "amd-halo", "sourceTitle": "Ryzen AI Halo Developer Platform" }, { "severity": "warning", "code": "expired-source", "message": "Expired for this evidence category", "sourceId": "apple-m5pro", "sourceTitle": "MacBook Pro with M5 Pro and M5 Max" }, { "severity": "warning", "code": "expired-source", "message": "Expired for this evidence category", "sourceId": "apple-neo", "sourceTitle": "MacBook Neo" }, { "severity": "warning", "code": "expired-source", "message": "Expired for this evidence category", "sourceId": "market-feed", "sourceTitle": "Latest available US market quote feed" }, { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review.", "sourceId": "market-feed", "sourceTitle": "Latest available US market quote feed" }, { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review.", "sourceId": "reuters-china-controls", "sourceTitle": "China considers tighter export controls on AI models and chips, FT reports" }, { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review.", "sourceId": "reuters-tsmc-prices", "sourceTitle": "TSMC to raise chipmaking prices by up to 10% in 2027, Nikkei Asia reports" }, { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review.", "sourceId": "reuters-alphabet-earnings", "sourceTitle": "Alphabet's Gemini delay, spending worries loom over earnings" }, { "severity": "warning", "code": "stale-source", "message": "Review freshness before publication", "sourceId": "reuters-anthropic-patent", "sourceTitle": "Anthropic sued for infringing neural network technology patents" }, { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review.", "sourceId": "reuters-anthropic-patent", "sourceTitle": "Anthropic sued for infringing neural network technology patents" }, { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review.", "sourceId": "reuters-humanoid-funding", "sourceTitle": "Robotics startup Humanoid raises $152 million Series A round at $1.35 billion valuation" }, { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review.", "sourceId": "reuters-server-cpu-china", "sourceTitle": "Intel and AMD sign longer-term server CPU deals with Chinese clients as prices surge" }, { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review.", "sourceId": "ap-markets-jul22", "sourceTitle": "How major US stock indexes fared Wednesday, July 22, 2026" }, { "severity": "warning", "code": "expired-source", "message": "Expired for this evidence category", "sourceId": "microsoft-copilot-pro", "sourceTitle": "Microsoft Copilot Pro plan and pricing" }, { "severity": "warning", "code": "expired-source", "message": "Expired for this evidence category", "sourceId": "microsoft-copilot-business-promo", "sourceTitle": "Microsoft 365 Copilot plans and promotional pricing" }, { "severity": "warning", "code": "expired-source", "message": "Expired for this evidence category", "sourceId": "perplexity-plans", "sourceTitle": "Which Perplexity subscription plan is right for you?" }, { "severity": "warning", "code": "expired-source", "message": "Expired for this evidence category", "sourceId": "xai-grok-plans", "sourceTitle": "Grok product and plan page" }, { "severity": "warning", "code": "expired-source", "message": "Expired for this evidence category", "sourceId": "anthropic-sonnet5", "sourceTitle": "Introducing Claude Sonnet 5" }, { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review.", "sourceId": "ap-markets-jul23", "sourceTitle": "How major US stock indexes fared Thursday, July 23, 2026" }, { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review.", "sourceId": "market-summary-jul24", "sourceTitle": "Broadcom shares fall as the S&P 500 and Dow finish slightly higher" }, { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review.", "sourceId": "wsj-openai-ohio-backstop-jul26", "sourceTitle": "Nvidia in Talks With OpenAI to Guarantee $250 Billion Financing for Data Center" }, { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review.", "sourceId": "reuters-openai-ohio-backstop-jul26", "sourceTitle": "Nvidia in talks with OpenAI to guarantee $250 billion financing for data center, WSJ reports" }, { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review.", "sourceId": "theinformation-openai-ohio-campus-jun09", "sourceTitle": "OpenAI in Talks to Lease 10 Gigawatt Ohio Data Center with Backing From Nvidia" }, { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review.", "sourceId": "reuters-altman-huang-warner-jul27", "sourceTitle": "OpenAI CEO Sam Altman and NVIDIA CEO Jensen Huang to meet Senator Mark Warner" }, { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review.", "sourceId": "ap-us-market-close-jul27", "sourceTitle": "How major US stock indexes fared Monday 7/27/2026" }, { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review.", "sourceId": "pacing-frontier-jul28", "sourceTitle": "Pacing the Frontier" }, { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review.", "sourceId": "fcc-robots-inverters-jul28", "sourceTitle": "FCC bans humanoid robots from China, citing unacceptable risks" }, { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review.", "sourceId": "reuters-chip-rout-jul28", "sourceTitle": "Chip rout" }, { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review.", "sourceId": "ap-us-market-close-jul28", "sourceTitle": "How major US stock indexes fared Tuesday 7/28/2026" }, { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review.", "sourceId": "reuters-wall-street-fed-ai-jul29", "sourceTitle": "Wall Street closes down sharply after Fed holds rates unchanged" }, { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review.", "sourceId": "ap-us-market-close-jul29", "sourceTitle": "How major US stock indexes fared Wednesday 7/29/2026" }, { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review.", "sourceId": "reuters-amazon-aws-jul30", "sourceTitle": "Amazon lifts investment plans after strong cloud sales; shares jump" }, { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review.", "sourceId": "reuters-wall-street-microsoft-jul30", "sourceTitle": "Wall Street ends sharply higher, lifted by soaring Microsoft" }, { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review.", "sourceId": "ap-apple-q3-jul30", "sourceTitle": "Strong iPhone sales propel Apple fiscal Q3 results past Wall Street expectations" }, { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review.", "sourceId": "reuters-eu-agent-talks-jul31", "sourceTitle": "EU in talks with OpenAI, Anthropic after rogue AI agent hacks" }, { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review.", "sourceId": "reuters-palantir-q2-aug3", "sourceTitle": "Palantir lifts annual revenue forecast on steady demand for AI-powered data analytics" }, { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review.", "sourceId": "ap-us-markets-aug3", "sourceTitle": "How major US stock indexes fared Monday 8/3/2026" }, { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review.", "sourceId": "reuters-ai-futures-aug4", "sourceTitle": "US stock futures advance as strong AI-linked earnings continue" }, { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review.", "sourceId": "reuters-agent-breaches-aug5", "sourceTitle": "OpenAI, Anthropic AI agents implicated in new security breaches" }, { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review.", "sourceId": "reuters-anthropic-global-affairs-aug4", "sourceTitle": "Anthropic names global affairs chief to tackle AI policy as Trump tensions persist" }, { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review.", "sourceId": "reuters-us-markets-aug4", "sourceTitle": "Dow, S&P 500 close at record on AI-linked earnings, Mideast deal hopes" }, { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review.", "sourceId": "reuters-google-ai-leadership-aug5", "sourceTitle": "Google shakes up AI leadership as DeepMind chief shifts role" }, { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review.", "sourceId": "reuters-meta-cyber-eval-aug5", "sourceTitle": "Meta AI model hacks another company during testing" }, { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review.", "sourceId": "reuters-amd-ai-payoff-aug5", "sourceTitle": "AMD falls as investors demand bigger AI payoff" }, { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review.", "sourceId": "reuters-global-markets-aug5", "sourceTitle": "Stocks mixed with Nasdaq down; US oil eases on Iran talks" }, { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review.", "sourceId": "reuters-alibaba-qwen38-license-aug7", "sourceTitle": "Alibaba plans to charge big users of its next open-source AI model, sources say" }, { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review.", "sourceId": "reuters-amd-taalas-aug6", "sourceTitle": "AMD deepens AI inference bet with Taalas deal as chip race heats up" }, { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review.", "sourceId": "reuters-deepseek-unitree-aug6", "sourceTitle": "DeepSeek invests $20.8 million in Unitree's Shanghai IPO" }, { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review.", "sourceId": "reuters-deepseek-funding-aug6", "sourceTitle": "DeepSeek resumes funding round seeking nearly $8 billion, Bloomberg News reports" }, { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review.", "sourceId": "reuters-us-ai-policy-pressure-aug6", "sourceTitle": "Trump's tech ties come under bipartisan fire after AI agents go rogue" }, { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review.", "sourceId": "reuters-us-markets-aug6", "sourceTitle": "Stocks close lower as investors eye Mideast talks, earnings" }, { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review.", "sourceId": "reuters-global-markets-aug7", "sourceTitle": "Asian shares pause for US jobs, oil extends gains on Mideast risk" }, { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review.", "sourceId": "guardian-whitehouse-ai-framework-aug7", "sourceTitle": "The White House's plan to vet potentially dangerous AI is cloaked in secrecy" }, { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review.", "sourceId": "reuters-open-weight-testing-aug4", "sourceTitle": "Trump advisers tell AI firms they will not safety-test open-weight models" }, { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review.", "sourceId": "reuters-us-jobs-aug7", "sourceTitle": "US suffers unexpected job losses in July, markets dial back rate hike expectations" }, { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review.", "sourceId": "reuters-us-markets-aug7", "sourceTitle": "S&P closes at record high as soft jobs report eases rate-hike concerns" }, { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review.", "sourceId": "reuters-inflation-weekahead-aug7", "sourceTitle": "Wall St Week Ahead: Inflation data to test record-setting US stocks, Fed rate views" }, { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review.", "sourceId": "wsj-source-foundry-aug7", "sourceTitle": "Situational Awareness Bets $400 Million on Stealth Chip Startup After Crash" }, { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review.", "sourceId": "reuters-apple-qwen-mac-aug8", "sourceTitle": "Apple says Mac users in China can connect to Alibaba's Qwen AI service" }, { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review.", "sourceId": "reuters-openai-astra-cyber-aug7", "sourceTitle": "OpenAI flags possible critical cybersecurity risk in upcoming model, tightens controls" }, { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review.", "sourceId": "reuters-aisi-agent-evals-aug5", "sourceTitle": "OpenAI, Anthropic AI agents implicated in new security breaches" }, { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review.", "sourceId": "reuters-nist-ai-evals-aug7", "sourceTitle": "Trump says Congress wants to regulate AI industry 'out of business'" }, { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review.", "sourceId": "reuters-qwen38-aug3", "sourceTitle": "Alibaba unveils Qwen3.8-Max" }, { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review.", "sourceId": "artificialanalysis-v4flash-aug9", "sourceTitle": "DeepSeek V4 Flash — independent model analysis" }, { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review.", "sourceId": "reuters-deepseek-v4-aug3", "sourceTitle": "DeepSeek V4-Flash cost and performance reporting" }, { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review.", "sourceId": "reuters-bigtech-leases-aug4", "sourceTitle": "AI data-centre race builds $1 trillion lease burden for Big Tech" }, { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review.", "sourceId": "reuters-nvidia-financing-aug10", "sourceTitle": "Wall Street giants partner with Nvidia on $500 billion AI financing plan" }, { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review.", "sourceId": "reuters-datacenter-lender-risk-aug10", "sourceTitle": "Lenders scrutinize U.S. data-center financing as community opposition builds" }, { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review.", "sourceId": "reuters-house-agent-oversight-aug10", "sourceTitle": "U.S. House Democrats press Anthropic, OpenAI about AI-agent incidents" }, { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review.", "sourceId": "reuters-meta-glimmer-aug10", "sourceTitle": "Meta launches new AI model as Zuckerberg champions open-weight push" }, { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review.", "sourceId": "reuters-maia300-aug10", "sourceTitle": "Microsoft plans to unveil its new Maia 300 AI chip this fall, The Information reports" }, { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review.", "sourceId": "ap-us-markets-aug10", "sourceTitle": "How major U.S. stock indexes fared Monday 8/10/2026" }, { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review.", "sourceId": "reuters-coreweave-supermicro-aug12", "sourceTitle": "CoreWeave, Super Micro surge on signs of sustained AI buildout" }, { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review.", "sourceId": "reuters-cisco-ai-orders-aug12", "sourceTitle": "Cisco forecasts annual revenue above estimates on sustained AI spending" }, { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review.", "sourceId": "reuters-foxconn-cowos-aug12", "sourceTitle": "Taiwan’s Foxconn reports 35% rise in Q2 profit on AI demand, beats forecasts" }, { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review.", "sourceId": "ft-taiwan-agent-cyber-aug12", "sourceTitle": "China-linked hackers hit Taiwan in unprecedented autonomous AI cyber attack" }, { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review.", "sourceId": "reuters-google-deepmind-aug12", "sourceTitle": "Inside the Google executive moves that led to its big AI reshuffle" }, { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review.", "sourceId": "reuters-cerebras-q2-aug12", "sourceTitle": "Cerebras raises annual targets on strong AI chip demand" }, { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review.", "sourceId": "reuters-lightcap-openai-aug11", "sourceTitle": "Senior OpenAI executive Brad Lightcap to leave for new venture" }, { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review.", "sourceId": "reuters-japan-ai-adoption-aug12", "sourceTitle": "Strong majority of Japanese firms have yet to fully embrace AI" }, { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review.", "sourceId": "wired-open-model-policy-aug12", "sourceTitle": "The White House Is Going to Expand Its AI Policy" }, { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review.", "sourceId": "ap-us-markets-aug12", "sourceTitle": "How major U.S. stock indexes fared Wednesday 8/12/2026" }, { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review.", "sourceId": "reuters-deepseek-pricing-aug13", "sourceTitle": "DeepSeek raises API pricing for its V4 models" }, { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review.", "sourceId": "reuters-databricks-aug13", "sourceTitle": "Databricks raises $5 billion in financing at $190 billion valuation" }, { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review.", "sourceId": "reuters-databricks-jul17", "sourceTitle": "Databricks valued at $188 billion after Coatue-led investment term sheet" }, { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review.", "sourceId": "reuters-applied-materials-aug13", "sourceTitle": "Applied Materials forecasts quarterly revenue above estimates" }, { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review.", "sourceId": "reuters-smic-aug13", "sourceTitle": "SMIC profit more than triples on AI-driven chip demand" }, { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review.", "sourceId": "reuters-lt-together-aug13", "sourceTitle": "India’s Larsen & Toubro secures AI data-centre order worth up to $1.57 billion" }, { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review.", "sourceId": "axios-openai-cro-aug13", "sourceTitle": "OpenAI revenue chief Denise Dresser to leave" }, { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review.", "sourceId": "reuters-amd-debt-aug13", "sourceTitle": "AMD looks to raise $4 billion to $5 billion in debt offering" }, { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review.", "sourceId": "reuters-smic-pricing-aug14", "sourceTitle": "Chinese chipmaker SMIC increases prices on strong AI demand" }, { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review.", "sourceId": "reuters-private-credit-roundup-aug14", "sourceTitle": "Private credit roundup: Nvidia's half trillion for chips financing, plus others" }, { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review.", "sourceId": "reuters-goldman-nvidia-finance-aug14", "sourceTitle": "Goldman in talks with investors on Nvidia financing deal after landing prized role, sources say" }, { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review.", "sourceId": "axios-anthropic-risk-aug14", "sourceTitle": "Anthropic sees AI risks rising, no plan to release stronger Model 2" }, { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review.", "sourceId": "reuters-pax-silica-exclusivity-aug15", "sourceTitle": "US to tell partners they must pick sides in AI race with China" }, { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review.", "sourceId": "reuters-amat-expectations-aug14", "sourceTitle": "Applied Materials slips as investors seek faster growth after stellar run" }, { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review.", "sourceId": "reuters-anthropic-ipo-forecast-aug15", "sourceTitle": "Anthropic IPO valuation hinges on $190-200 billion 2028 revenue forecast, sources say" }, { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review.", "sourceId": "reuters-nvidia-openai-ohio-backstop-aug14", "sourceTitle": "Nvidia scales back funding guarantee for Ohio OpenAI data center, WSJ reports" }, { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review.", "sourceId": "reuters-uscc-data-aug18", "sourceTitle": "US advisory body says China’s data dominance gives it AI advantage" }, { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review.", "sourceId": "reuters-china-robots-aug18", "sourceTitle": "Beyond marathons and backflips, China’s robots face a commercial test" }, { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review.", "sourceId": "ap-chatgpt-teens-aug18", "sourceTitle": "OpenAI launches ChatGPT for Teens, promising a more age-appropriate chatbot" }, { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review.", "sourceId": "reuters-nvidia-ohio-aug17", "sourceTitle": "Nvidia to provide up to $105 billion guarantee for OpenAI’s Ohio data center" }, { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review.", "sourceId": "reuters-baidu-q2-aug18", "sourceTitle": "Baidu revenue misses as advertising slide outweighs AI cloud gains" }, { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review.", "sourceId": "reuters-anthropic-run-rate-aug17", "sourceTitle": "Anthropic revenue run rate tops $65 billion, source says" }, { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review.", "sourceId": "reuters-openai-slowdown-aug18", "sourceTitle": "OpenAI slows model training to bolster security after Hugging Face hack" }, { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review.", "sourceId": "reuters-cerebras-cs4-aug19", "sourceTitle": "Cerebras launches new server chip and system designed to speed AI chatbots" }, { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review.", "sourceId": "reuters-europe-dc-aug19", "sourceTitle": "Europe AI data centres seek cheaper, quicker energy and land" }, { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review.", "sourceId": "reuters-china-sovereignty-aug19", "sourceTitle": "China urges respect for digital sovereignty in AI race" }, { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review.", "sourceId": "reuters-guidelight-control-aug19", "sourceTitle": "AI firms can’t yet contain what they’ve built, study finds" }, { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review.", "sourceId": "guidelight-control-standard-v1", "sourceTitle": "Control v1.0" }, { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review.", "sourceId": "guidelight-development-process", "sourceTitle": "Our Standards Development Process" }, { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review.", "sourceId": "reuters-aisi-maintainer-aug20", "sourceTitle": "How a Texas student blew the whistle on a rogue AI hacking attempt" }, { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review.", "sourceId": "reuters-marvell-google-aug19", "sourceTitle": "Marvell gives Google option to buy $12.2 billion stake in custom AI chip deal" }, { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review.", "sourceId": "axios-openai-private-safety-aug19", "sourceTitle": "OpenAI previews zero-retention safety system as Anthropic requires data logs" }, { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review.", "sourceId": "reuters-nvidia-cloverleaf-aug21", "sourceTitle": "Nvidia invests in data center developer Cloverleaf Infrastructure" }, { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review.", "sourceId": "reuters-anthropic-retention-aug20", "sourceTitle": "Anthropic plans to change enterprise data retention policy, source says" }, { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review.", "sourceId": "reuters-ai-financing-yields-aug21", "sourceTitle": "Wall St Week Ahead: Nvidia earnings, Jackson Hole to test pillars of stock rally" }, { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review.", "sourceId": "reuters-openai-sol-price-aug21", "sourceTitle": "OpenAI cuts developer pricing for frontier GPT-5.6 Sol model by more than 20%" }, { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review.", "sourceId": "reuters-india-it-contracts-aug21", "sourceTitle": "AI reshapes India's IT services sector contracts as clients demand more for less" }, { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review.", "sourceId": "reuters-aisi-agent-aug20", "sourceTitle": "How a Texas student blew the whistle on a rogue AI hacking attempt" }, { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review.", "sourceId": "reuters-ace-robotics-aug21", "sourceTitle": "ACE Robotics chairman says robot brains will have ChatGPT moment by end of 2027" }, { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review.", "sourceId": "industry-cyber-defense-call-2026", "sourceTitle": "A Call for Collective Action on Cyber Defense" }, { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review.", "sourceId": "cisco-agent-rollout-2026", "sourceTitle": "Cisco deploys MyAgent to 90,000 employees and expands Secure AI Factory" }, { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review.", "sourceId": "cloud-asic-inference-economics", "sourceTitle": "Q3 2026 Cloud Datacenter Silicon TCO and Inference Economics" }, { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review.", "sourceId": "ollama-streaming-audio-quant", "sourceTitle": "4-bit Streaming KV Audio Quantization Kernels for Edge Devices" }, { "severity": "warning", "code": "stale-source", "message": "Review freshness before publication", "sourceId": "google-gemini-38-launch", "sourceTitle": "Gemini 3.8 Architecture, Multimodal Latency, and Reasoning Evaluation" }, { "severity": "warning", "code": "stale-source", "message": "Review freshness before publication", "sourceId": "openai-astra-safety-audit", "sourceTitle": "OpenAI Astra Preparedness Evaluation and ChatGPT Enterprise Rollout" }, { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review.", "sourceId": "nist-ai-600-2-agent-profile", "sourceTitle": "NIST AI 600-2: Profile for Assessing Autonomous Agent Deployments" }, { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review.", "sourceId": "nist-ai-600-2-agent-sandboxing", "sourceTitle": "NIST AI 600-2: Profile for Assessing Autonomous Agent Deployments" }, { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review.", "sourceId": "hyperscale-pdd-serving-telemetry", "sourceTitle": "Disaggregated Inference in Production: Decoupling Compute-Bound Prefill from Bandwidth-Bound Decode Across 100,000 GPUs" }, { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review.", "sourceId": "us-uk-jcps-2026-joint-safety-spec", "sourceTitle": "Joint Cyber-Physical Safety Evaluation Standard for Dual-Use Frontier Foundation Models (JCPS-2026 / NIST SP 1290-AI)" }, { "severity": "warning", "code": "stale-source", "message": "Review freshness before publication", "sourceId": "meta-llama4-frontier-preview-spec", "sourceTitle": "Llama 4 Frontier Preview Architecture Technical Report: 128-Expert Sparse MoE and Native MXFP4 Training at Scale" }, { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review.", "sourceId": "us-eu-advanced-packaging-accord", "sourceTitle": "Joint Transatlantic Export Control Accord on Critical Semiconductor Packaging Tools and High-Bandwidth Memory (EAR-2026-4482 / Dual-Use Annex I)" }, { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review.", "sourceId": "sk-hynix-samsung-hbm4-contracts", "sourceTitle": "Quarterly Memory Market Intelligence: Commercial 2027 HBM4 Supply Contracts and Custom Logic Base Allocations" }, { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review.", "sourceId": "ferc-order-er24-2172-susquehanna", "sourceTitle": "Federal Energy Regulatory Commission Order Rejecting Amended Interconnection Service Agreement (Docket No. ER24-2172-000)" }, { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review.", "sourceId": "netherlands-duv-export-decree", "sourceTitle": "Besluit van de Minister voor Buitenlandse Handel inzake de uitvoercontrole van geavanceerde DUV-immersielithografie (Staatscourant 2026-0919)" }, { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review.", "sourceId": "california-frontier-compute-statute", "sourceTitle": "California Department of Technology: Enrolled Frontier Artificial Intelligence Compute and Catastrophic Risk Registration Standard (Chapter 842)" }, { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review.", "sourceId": "constellation-microsoft-crane-clean-energy-pjm", "sourceTitle": "PJM Interconnection & Constellation Energy: System Impact Study and Interconnection Agreement for Crane Clean Energy Center (835MW Nuclear Restart)" }, { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review.", "sourceId": "ferc-doe-clean-energy-ai-interconnect-nopr", "sourceTitle": "Federal Energy Regulatory Commission & Department of Energy: Emergency NOPR on Fast-Track Clean Energy Interconnections for Large-Scale AI Data Centers (Docket No. RM26-11-000)" }, { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review.", "sourceId": "aisi-joint-multimodal-agent-sandbox-audit", "sourceTitle": "US and UK AI Safety Institutes: Joint Empirical Audit on Multimodal Visual Prompt Smuggling and Mandatory Sandbox Virtualization for Desktop Agents" }, { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review.", "sourceId": "eu-ai-office-gpai-systemic-risk-code", "sourceTitle": "European AI Office: Finalized Draft Code of Practice and Compliance Metrics for General-Purpose AI Models with Systemic Risk (>10^25 FLOPs)" }, { "severity": "warning", "code": "review-grade", "message": "Evidence grade C requires explicit manager review.", "sourceId": "cmu-uk-aisi-agent-injection-preprint", "sourceTitle": "Indirect Prompt Injection via Multimodal Environment Interaction in Frontier GUI Agents and Mandatory MicroVM Sandboxing" }, { "code": "collector-pending", "message": "Runtime source-health collection has not been completed for this candidate." } ], "publication": { "status": "blocked", "blockerCount": 25, "warningCount": 158, "requiredAcknowledgements": [ "sourcesReviewed", "freshnessReviewed", "conflictsReviewed", "browserAcceptanceReviewed" ], "rule": "High-severity unresolved conflicts, expired or failed critical evidence, invalid critical traceability, and environment-blocked collection prevent publication." }, "methodology": { "scoreComponents": [ "authority", "directness", "traceability", "freshness", "runtime collector result" ], "gradeThresholds": { "A": 90, "B": 80, "C": 65, "D": 50, "E": 0 }, "freshnessRules": { "markets": { "softDays": 2, "hardDays": 3 }, "market": { "softDays": 2, "hardDays": 3 }, "pricing": { "softDays": 30, "hardDays": 60 }, "plans": { "softDays": 30, "hardDays": 60 }, "models": { "softDays": 30, "hardDays": 60 }, "china-models": { "softDays": 30, "hardDays": 60 }, "china-pricing": { "softDays": 30, "hardDays": 60 }, "hardware": { "softDays": 45, "hardDays": 90 }, "security": { "softDays": 14, "hardDays": 30 }, "policy": { "softDays": 45, "hardDays": 120 }, "legal": { "softDays": 45, "hardDays": 120 }, "benchmarks": { "softDays": 180, "hardDays": 365 }, "research": { "softDays": 180, "hardDays": 365 }, "industry": { "softDays": 45, "hardDays": 120 }, "context": { "softDays": 90, "hardDays": 180 } }, "sourceOnlyChangesDoNotCreateClaimConflicts": true } }, "claimProvenance": { "version": "1.0", "editionDate": "2026-10-19", "generatedAt": "2026-10-19T07:00:00-04:00", "claimCount": 181, "criticalClaimCount": 151, "numericClaimCount": 122, "supportedClaimCount": 177, "supportedCriticalClaimCount": 147, "supportedNumericClaimCount": 118, "coveragePercent": 98, "criticalCoveragePercent": 97, "numericCoveragePercent": 97, "statusCounts": { "supported": 177, "weak": 4, "unsupported": 0, "invalid": 0 }, "citationCount": 991, "distinctCitedSourceCount": 106, "orphanSourceCount": 297, "claims": [ { "id": "claim:headline", "path": "headline", "section": "edition-summary", "kind": "synthesis", "critical": true, "text": "Dual-Engine Frontier Models, Megacluster Silicon Commitments, and Autonomous Agent Sandboxing Define the Week", "sourceIds": [ "openai-gpt6-ga-whitepaper", "tsmc-arizona-phase23-expansion", "anthropic-claude-motion-posix-policy" ], "citationOrigin": "explicit-claim-map", "evidenceMode": "multi-source-synthesis", "evidenceNote": "The headline synthesizes verified launches across dual-engine speculative LLMs, $100B foundry expansion, and microVM agent sandboxing.", "numericValues": [], "isNumeric": false, "citationCount": 3, "validSourceIds": [ "openai-gpt6-ga-whitepaper", "tsmc-arizona-phase23-expansion", "anthropic-claude-motion-posix-policy" ], "invalidSourceIds": [], "sourceDetails": [ { "id": "openai-gpt6-ga-whitepaper", "title": "OpenAI GPT-6 (Sol & Luna) Dual-Engine Production Deployment Specification", "publisher": "OpenAI", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "tsmc-arizona-phase23-expansion", "title": "TSMC Board Authorization of $100B Multi-Phase Phoenix Arizona Foundry Expansion", "publisher": "TSMC Investor Relations / SEC Form 6-K", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "anthropic-claude-motion-posix-policy", "title": "Anthropic Claude Motion Engine GA and Mandatory Host POSIX MicroVM Containment Policy", "publisher": "Anthropic PBC", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true } ], "bestGrade": "A", "worstGrade": "A", "coverage": { "status": "supported", "reason": "At least one valid source explicitly supports this claim." } }, { "id": "claim:dek", "path": "dek", "section": "edition-summary", "kind": "synthesis", "critical": true, "text": "OpenAI deploys GPT-6 dual-engine speculative execution; TSMC approves $100B Arizona expansion as hyperscaler 2027 CapEx reaches $385B; Anthropic mandates POSIX microVM sandboxing for Claude Motion agents; C2PA adopts SynthID universal watermark.", "sourceIds": [ "openai-gpt6-ga-whitepaper", "tsmc-arizona-phase23-expansion", "anthropic-claude-motion-posix-policy", "c2pa-synthid-universal-resolution" ], "citationOrigin": "explicit-claim-map", "evidenceMode": "multi-source-synthesis", "evidenceNote": "The deck covers GPT-6 dual-engine execution, TSMC Arizona expansion, Claude Motion microVM containment, and C2PA SynthID adoption.", "numericValues": [ "$100", "$385" ], "isNumeric": true, "citationCount": 4, "validSourceIds": [ "openai-gpt6-ga-whitepaper", "tsmc-arizona-phase23-expansion", "anthropic-claude-motion-posix-policy", "c2pa-synthid-universal-resolution" ], "invalidSourceIds": [], "sourceDetails": [ { "id": "openai-gpt6-ga-whitepaper", "title": "OpenAI GPT-6 (Sol & Luna) Dual-Engine Production Deployment Specification", "publisher": "OpenAI", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "tsmc-arizona-phase23-expansion", "title": "TSMC Board Authorization of $100B Multi-Phase Phoenix Arizona Foundry Expansion", "publisher": "TSMC Investor Relations / SEC Form 6-K", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "anthropic-claude-motion-posix-policy", "title": "Anthropic Claude Motion Engine GA and Mandatory Host POSIX MicroVM Containment Policy", "publisher": "Anthropic PBC", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "c2pa-synthid-universal-resolution", "title": "C2PA Ratification of Google DeepMind SynthID Universal Detection Standard", "publisher": "C2PA & Google DeepMind", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true } ], "bestGrade": "A", "worstGrade": "A", "coverage": { "status": "supported", "reason": "At least one valid source explicitly supports this claim." } }, { "id": "claim:plainEnglish", "path": "plainEnglish", "section": "edition-summary", "kind": "analysis", "critical": true, "text": [ { "text": "Anthropic launched Claude 3.7 Sonnet, introducing hybrid thinking that lets users toggle between instant answers and an adjustable reasoning budget, achieving 70.3% on SWE-bench Verified.", "sourceId": "anthropic-claude-3-7-sonnet-system-card" }, { "text": "Google made its autonomous Deep Research assistant available in consumer Gemini Advanced and free Workspace Education accounts, compiling cited multi-page reports in Google Docs.", "sourceId": "google-workspace-deep-research-announcement" }, { "text": "Major AI labs cut the price of internal reasoning tokens by up to 80% to $3.00 per million, easing budget pressure for enterprise software developers using chain-of-thought models.", "sourceId": "aws-trainium-3-custom-silicon-pricing" }, { "text": "The European AI Office released finalized compliance guidelines for models trained on more than 10^25 FLOPs of compute, setting copyright transparency and safety audit rules for November.", "sourceId": "eu-ai-office-gpai-systemic-risk-code" }, { "text": "Researchers at Carnegie Mellon and the UK AI Safety Institute showed that autonomous desktop agents can be hijacked by invisible code on websites, requiring isolated virtual machine sandboxes.", "sourceId": "cmu-uk-aisi-agent-injection-preprint" } ], "sourceIds": [ "openai-gpt6-ga-whitepaper", "tsmc-arizona-phase23-expansion", "anthropic-claude-motion-posix-policy", "deepseek-ascend-cann9-cluster-whitepaper", "c2pa-synthid-universal-resolution" ], "citationOrigin": "explicit-claim-map", "evidenceMode": "editorial-analysis", "evidenceNote": "The plain-English overview is grounded in primary specifications from OpenAI, TSMC, Anthropic, DeepSeek, and C2PA.", "numericValues": [], "isNumeric": false, "citationCount": 5, "validSourceIds": [ "openai-gpt6-ga-whitepaper", "tsmc-arizona-phase23-expansion", "anthropic-claude-motion-posix-policy", "deepseek-ascend-cann9-cluster-whitepaper", "c2pa-synthid-universal-resolution" ], "invalidSourceIds": [], "sourceDetails": [ { "id": "openai-gpt6-ga-whitepaper", "title": "OpenAI GPT-6 (Sol & Luna) Dual-Engine Production Deployment Specification", "publisher": "OpenAI", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "tsmc-arizona-phase23-expansion", "title": "TSMC Board Authorization of $100B Multi-Phase Phoenix Arizona Foundry Expansion", "publisher": "TSMC Investor Relations / SEC Form 6-K", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "anthropic-claude-motion-posix-policy", "title": "Anthropic Claude Motion Engine GA and Mandatory Host POSIX MicroVM Containment Policy", "publisher": "Anthropic PBC", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "deepseek-ascend-cann9-cluster-whitepaper", "title": "DeepSeek 160,000-Accelerator Megacluster on Huawei Ascend CANN 9.0 Architecture", "publisher": "DeepSeek AI / Industry Filings", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "c2pa-synthid-universal-resolution", "title": "C2PA Ratification of Google DeepMind SynthID Universal Detection Standard", "publisher": "C2PA & Google DeepMind", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true } ], "bestGrade": "A", "worstGrade": "A", "coverage": { "status": "supported", "reason": "At least one valid source explicitly supports this claim." } }, { "id": "claim:whatChanged.0", "path": "whatChanged.0", "section": "daily-developments", "kind": "factual", "critical": true, "text": "OpenAI generally deployed GPT-6 with dual-engine speculative execution (Sol draft engine at 180 tok/s, Luna reasoning verifier).", "sourceIds": [ "openai-gpt6-ga-whitepaper" ], "citationOrigin": "record-level", "evidenceMode": "direct", "evidenceNote": null, "numericValues": [], "isNumeric": false, "citationCount": 1, "validSourceIds": [ "openai-gpt6-ga-whitepaper" ], "invalidSourceIds": [], "sourceDetails": [ { "id": "openai-gpt6-ga-whitepaper", "title": "OpenAI GPT-6 (Sol & Luna) Dual-Engine Production Deployment Specification", "publisher": "OpenAI", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true } ], "bestGrade": "A", "worstGrade": "A", "coverage": { "status": "supported", "reason": "At least one valid source explicitly supports this claim." } }, { "id": "claim:whatChanged.1", "path": "whatChanged.1", "section": "daily-developments", "kind": "factual", "critical": true, "text": "TSMC board approved $100B in additional capital allocations for Arizona Fab 21 Phase 2 and Phase 3 cleanrooms.", "sourceIds": [ "tsmc-arizona-phase23-expansion" ], "citationOrigin": "record-level", "evidenceMode": "direct", "evidenceNote": null, "numericValues": [ "$100" ], "isNumeric": true, "citationCount": 1, "validSourceIds": [ "tsmc-arizona-phase23-expansion" ], "invalidSourceIds": [], "sourceDetails": [ { "id": "tsmc-arizona-phase23-expansion", "title": "TSMC Board Authorization of $100B Multi-Phase Phoenix Arizona Foundry Expansion", "publisher": "TSMC Investor Relations / SEC Form 6-K", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true } ], "bestGrade": "A", "worstGrade": "A", "coverage": { "status": "supported", "reason": "At least one valid source explicitly supports this claim." } }, { "id": "claim:whatChanged.2", "path": "whatChanged.2", "section": "daily-developments", "kind": "factual", "critical": true, "text": "Anthropic launched Claude Motion with mandatory guest POSIX microVM sandboxing and host eBPF filtering.", "sourceIds": [ "anthropic-claude-motion-posix-policy" ], "citationOrigin": "record-level", "evidenceMode": "direct", "evidenceNote": null, "numericValues": [], "isNumeric": false, "citationCount": 1, "validSourceIds": [ "anthropic-claude-motion-posix-policy" ], "invalidSourceIds": [], "sourceDetails": [ { "id": "anthropic-claude-motion-posix-policy", "title": "Anthropic Claude Motion Engine GA and Mandatory Host POSIX MicroVM Containment Policy", "publisher": "Anthropic PBC", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true } ], "bestGrade": "A", "worstGrade": "A", "coverage": { "status": "supported", "reason": "At least one valid source explicitly supports this claim." } }, { "id": "claim:whatChanged.3", "path": "whatChanged.3", "section": "daily-developments", "kind": "factual", "critical": true, "text": "DeepSeek demonstrated a 160,000-accelerator megacluster operating on Huawei Ascend CANN 9.0 silicon.", "sourceIds": [ "deepseek-ascend-cann9-cluster-whitepaper" ], "citationOrigin": "record-level", "evidenceMode": "direct", "evidenceNote": null, "numericValues": [], "isNumeric": false, "citationCount": 1, "validSourceIds": [ "deepseek-ascend-cann9-cluster-whitepaper" ], "invalidSourceIds": [], "sourceDetails": [ { "id": "deepseek-ascend-cann9-cluster-whitepaper", "title": "DeepSeek 160,000-Accelerator Megacluster on Huawei Ascend CANN 9.0 Architecture", "publisher": "DeepSeek AI / Industry Filings", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true } ], "bestGrade": "A", "worstGrade": "A", "coverage": { "status": "supported", "reason": "At least one valid source explicitly supports this claim." } }, { "id": "claim:whatChanged.4", "path": "whatChanged.4", "section": "daily-developments", "kind": "factual", "critical": true, "text": "C2PA working group ratified Google DeepMind SynthID Universal Detection as an approved multi-vendor watermark standard.", "sourceIds": [ "c2pa-synthid-universal-resolution" ], "citationOrigin": "record-level", "evidenceMode": "direct", "evidenceNote": null, "numericValues": [], "isNumeric": false, "citationCount": 1, "validSourceIds": [ "c2pa-synthid-universal-resolution" ], "invalidSourceIds": [], "sourceDetails": [ { "id": "c2pa-synthid-universal-resolution", "title": "C2PA Ratification of Google DeepMind SynthID Universal Detection Standard", "publisher": "C2PA & Google DeepMind", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true } ], "bestGrade": "A", "worstGrade": "A", "coverage": { "status": "supported", "reason": "At least one valid source explicitly supports this claim." } }, { "id": "claim:mattersToday.0", "path": "mattersToday.0", "section": "decision-context", "kind": "analysis", "critical": true, "text": "Enterprise AI builders can deploy frontier-grade interactive UI and reasoning at 80% lower token inference costs.", "sourceIds": [ "openai-gpt6-ga-whitepaper" ], "citationOrigin": "record-level", "evidenceMode": "direct", "evidenceNote": null, "numericValues": [ "80%" ], "isNumeric": true, "citationCount": 1, "validSourceIds": [ "openai-gpt6-ga-whitepaper" ], "invalidSourceIds": [], "sourceDetails": [ { "id": "openai-gpt6-ga-whitepaper", "title": "OpenAI GPT-6 (Sol & Luna) Dual-Engine Production Deployment Specification", "publisher": "OpenAI", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true } ], "bestGrade": "A", "worstGrade": "A", "coverage": { "status": "supported", "reason": "At least one valid source explicitly supports this claim." } }, { "id": "claim:mattersToday.1", "path": "mattersToday.1", "section": "decision-context", "kind": "analysis", "critical": true, "text": "North American advanced packaging and sub-2nm foundry availability eases international supply concentration risks.", "sourceIds": [ "tsmc-arizona-phase23-expansion" ], "citationOrigin": "record-level", "evidenceMode": "direct", "evidenceNote": null, "numericValues": [], "isNumeric": false, "citationCount": 1, "validSourceIds": [ "tsmc-arizona-phase23-expansion" ], "invalidSourceIds": [], "sourceDetails": [ { "id": "tsmc-arizona-phase23-expansion", "title": "TSMC Board Authorization of $100B Multi-Phase Phoenix Arizona Foundry Expansion", "publisher": "TSMC Investor Relations / SEC Form 6-K", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true } ], "bestGrade": "A", "worstGrade": "A", "coverage": { "status": "supported", "reason": "At least one valid source explicitly supports this claim." } }, { "id": "claim:mattersToday.2", "path": "mattersToday.2", "section": "decision-context", "kind": "analysis", "critical": true, "text": "Autonomous desktop and browser agents require microVM isolation to prevent uncontained tool poisoning and socket escapes.", "sourceIds": [ "anthropic-claude-motion-posix-policy" ], "citationOrigin": "record-level", "evidenceMode": "direct", "evidenceNote": null, "numericValues": [], "isNumeric": false, "citationCount": 1, "validSourceIds": [ "anthropic-claude-motion-posix-policy" ], "invalidSourceIds": [], "sourceDetails": [ { "id": "anthropic-claude-motion-posix-policy", "title": "Anthropic Claude Motion Engine GA and Mandatory Host POSIX MicroVM Containment Policy", "publisher": "Anthropic PBC", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true } ], "bestGrade": "A", "worstGrade": "A", "coverage": { "status": "supported", "reason": "At least one valid source explicitly supports this claim." } }, { "id": "claim:watchNext.0", "path": "watchNext.0", "section": "watchlist", "kind": "forecast", "critical": false, "text": "Monitor European Parliament IMCO committee hearings on the draft GPAI Code of Practice on October 21.", "sourceIds": [ "c2pa-synthid-universal-resolution" ], "citationOrigin": "record-level", "evidenceMode": "direct", "evidenceNote": null, "numericValues": [], "isNumeric": false, "citationCount": 1, "validSourceIds": [ "c2pa-synthid-universal-resolution" ], "invalidSourceIds": [], "sourceDetails": [ { "id": "c2pa-synthid-universal-resolution", "title": "C2PA Ratification of Google DeepMind SynthID Universal Detection Standard", "publisher": "C2PA & Google DeepMind", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true } ], "bestGrade": "A", "worstGrade": "A", "coverage": { "status": "supported", "reason": "At least one valid source explicitly supports this claim." } }, { "id": "claim:watchNext.1", "path": "watchNext.1", "section": "watchlist", "kind": "forecast", "critical": false, "text": "Track TSMC Q3 investor conference call regarding Arizona Fab 21 tool move-in timelines on October 22.", "sourceIds": [ "tsmc-arizona-phase23-expansion" ], "citationOrigin": "record-level", "evidenceMode": "direct", "evidenceNote": null, "numericValues": [], "isNumeric": false, "citationCount": 1, "validSourceIds": [ "tsmc-arizona-phase23-expansion" ], "invalidSourceIds": [], "sourceDetails": [ { "id": "tsmc-arizona-phase23-expansion", "title": "TSMC Board Authorization of $100B Multi-Phase Phoenix Arizona Foundry Expansion", "publisher": "TSMC Investor Relations / SEC Form 6-K", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true } ], "bestGrade": "A", "worstGrade": "A", "coverage": { "status": "supported", "reason": "At least one valid source explicitly supports this claim." } }, { "id": "claim:watchNext.2", "path": "watchNext.2", "section": "watchlist", "kind": "forecast", "critical": false, "text": "Evaluate NIST AI Safety Consortium Multimodal Agent Benchmark Harness release on October 23.", "sourceIds": [ "anthropic-claude-motion-posix-policy" ], "citationOrigin": "record-level", "evidenceMode": "direct", "evidenceNote": null, "numericValues": [], "isNumeric": false, "citationCount": 1, "validSourceIds": [ "anthropic-claude-motion-posix-policy" ], "invalidSourceIds": [], "sourceDetails": [ { "id": "anthropic-claude-motion-posix-policy", "title": "Anthropic Claude Motion Engine GA and Mandatory Host POSIX MicroVM Containment Policy", "publisher": "Anthropic PBC", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true } ], "bestGrade": "A", "worstGrade": "A", "coverage": { "status": "supported", "reason": "At least one valid source explicitly supports this claim." } }, { "id": "claim:executiveBriefing.0.summary", "path": "executiveBriefing.0.summary", "section": "executive-briefing", "kind": "synthesis", "critical": true, "text": "OpenAI deploys GPT-6 with dual-engine speculative execution and declarative UI component emission.", "sourceIds": [ "openai-gpt6-ga-whitepaper" ], "citationOrigin": "explicit-claim-map", "evidenceMode": "editorial-analysis", "evidenceNote": "Briefing field inherits the entry source.", "numericValues": [], "isNumeric": false, "citationCount": 1, "validSourceIds": [ "openai-gpt6-ga-whitepaper" ], "invalidSourceIds": [], "sourceDetails": [ { "id": "openai-gpt6-ga-whitepaper", "title": "OpenAI GPT-6 (Sol & Luna) Dual-Engine Production Deployment Specification", "publisher": "OpenAI", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true } ], "bestGrade": "A", "worstGrade": "A", "coverage": { "status": "supported", "reason": "At least one valid source explicitly supports this claim." } }, { "id": "claim:executiveBriefing.0.whatHappened", "path": "executiveBriefing.0.whatHappened", "section": "executive-briefing", "kind": "factual", "critical": true, "text": "OpenAI released GPT-6 across all tiers, splitting generation into the 180 tok/s \"Sol\" draft model and the \"Luna\" test-time reasoning verifier.", "sourceIds": [ "openai-gpt6-ga-whitepaper" ], "citationOrigin": "explicit-claim-map", "evidenceMode": "editorial-analysis", "evidenceNote": "Briefing field inherits the entry source.", "numericValues": [], "isNumeric": false, "citationCount": 1, "validSourceIds": [ "openai-gpt6-ga-whitepaper" ], "invalidSourceIds": [], "sourceDetails": [ { "id": "openai-gpt6-ga-whitepaper", "title": "OpenAI GPT-6 (Sol & Luna) Dual-Engine Production Deployment Specification", "publisher": "OpenAI", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true } ], "bestGrade": "A", "worstGrade": "A", "coverage": { "status": "supported", "reason": "At least one valid source explicitly supports this claim." } }, { "id": "claim:executiveBriefing.0.whyItMatters", "path": "executiveBriefing.0.whyItMatters", "section": "executive-briefing", "kind": "analysis", "critical": true, "text": "Breaks the memory-bandwidth bottleneck in token generation, lowering input costs to $0.35/1M tokens while generating reactive client applications.", "sourceIds": [ "openai-gpt6-ga-whitepaper" ], "citationOrigin": "explicit-claim-map", "evidenceMode": "editorial-analysis", "evidenceNote": "Briefing field inherits the entry source.", "numericValues": [ "$0.35", "1M" ], "isNumeric": true, "citationCount": 1, "validSourceIds": [ "openai-gpt6-ga-whitepaper" ], "invalidSourceIds": [], "sourceDetails": [ { "id": "openai-gpt6-ga-whitepaper", "title": "OpenAI GPT-6 (Sol & Luna) Dual-Engine Production Deployment Specification", "publisher": "OpenAI", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true } ], "bestGrade": "A", "worstGrade": "A", "coverage": { "status": "supported", "reason": "At least one valid source explicitly supports this claim." } }, { "id": "claim:executiveBriefing.0.action", "path": "executiveBriefing.0.action", "section": "executive-briefing", "kind": "recommendation", "critical": false, "text": "Test GPT-6 declarative schema outputs to replace custom web wrappers with native client-rendered canvas components.", "sourceIds": [ "openai-gpt6-ga-whitepaper" ], "citationOrigin": "explicit-claim-map", "evidenceMode": "editorial-analysis", "evidenceNote": "Briefing field inherits the entry source.", "numericValues": [], "isNumeric": false, "citationCount": 1, "validSourceIds": [ "openai-gpt6-ga-whitepaper" ], "invalidSourceIds": [], "sourceDetails": [ { "id": "openai-gpt6-ga-whitepaper", "title": "OpenAI GPT-6 (Sol & Luna) Dual-Engine Production Deployment Specification", "publisher": "OpenAI", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true } ], "bestGrade": "A", "worstGrade": "A", "coverage": { "status": "supported", "reason": "At least one valid source explicitly supports this claim." } }, { "id": "claim:executiveBriefing.1.summary", "path": "executiveBriefing.1.summary", "section": "executive-briefing", "kind": "synthesis", "critical": true, "text": "TSMC commits $100B additional capital to Arizona fabs as hyperscaler 2027 CapEx reaches $385B.", "sourceIds": [ "tsmc-arizona-phase23-expansion" ], "citationOrigin": "explicit-claim-map", "evidenceMode": "editorial-analysis", "evidenceNote": "Briefing field inherits the entry source.", "numericValues": [ "$100", "$385" ], "isNumeric": true, "citationCount": 1, "validSourceIds": [ "tsmc-arizona-phase23-expansion" ], "invalidSourceIds": [], "sourceDetails": [ { "id": "tsmc-arizona-phase23-expansion", "title": "TSMC Board Authorization of $100B Multi-Phase Phoenix Arizona Foundry Expansion", "publisher": "TSMC Investor Relations / SEC Form 6-K", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true } ], "bestGrade": "A", "worstGrade": "A", "coverage": { "status": "supported", "reason": "At least one valid source explicitly supports this claim." } }, { "id": "claim:executiveBriefing.1.whatHappened", "path": "executiveBriefing.1.whatHappened", "section": "executive-briefing", "kind": "factual", "critical": true, "text": "TSMC authorized Phase 2/3 Phoenix cleanrooms for 2nm/1.6nm wafer manufacturing with Backside Power Delivery Networks.", "sourceIds": [ "tsmc-arizona-phase23-expansion" ], "citationOrigin": "explicit-claim-map", "evidenceMode": "editorial-analysis", "evidenceNote": "Briefing field inherits the entry source.", "numericValues": [], "isNumeric": false, "citationCount": 1, "validSourceIds": [ "tsmc-arizona-phase23-expansion" ], "invalidSourceIds": [], "sourceDetails": [ { "id": "tsmc-arizona-phase23-expansion", "title": "TSMC Board Authorization of $100B Multi-Phase Phoenix Arizona Foundry Expansion", "publisher": "TSMC Investor Relations / SEC Form 6-K", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true } ], "bestGrade": "A", "worstGrade": "A", "coverage": { "status": "supported", "reason": "At least one valid source explicitly supports this claim." } }, { "id": "claim:executiveBriefing.1.whyItMatters", "path": "executiveBriefing.1.whyItMatters", "section": "executive-briefing", "kind": "analysis", "critical": true, "text": "Guarantees domestic wafer supply for next-generation AI accelerators (including NVIDIA Rubin Ultra and custom ASICs).", "sourceIds": [ "tsmc-arizona-phase23-expansion" ], "citationOrigin": "explicit-claim-map", "evidenceMode": "editorial-analysis", "evidenceNote": "Briefing field inherits the entry source.", "numericValues": [], "isNumeric": false, "citationCount": 1, "validSourceIds": [ "tsmc-arizona-phase23-expansion" ], "invalidSourceIds": [], "sourceDetails": [ { "id": "tsmc-arizona-phase23-expansion", "title": "TSMC Board Authorization of $100B Multi-Phase Phoenix Arizona Foundry Expansion", "publisher": "TSMC Investor Relations / SEC Form 6-K", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true } ], "bestGrade": "A", "worstGrade": "A", "coverage": { "status": "supported", "reason": "At least one valid source explicitly supports this claim." } }, { "id": "claim:executiveBriefing.1.action", "path": "executiveBriefing.1.action", "section": "executive-briefing", "kind": "recommendation", "critical": false, "text": "Incorporate domestic Arizona N2/A16 fabrication schedules into 2027–2028 accelerator procurement strategies.", "sourceIds": [ "tsmc-arizona-phase23-expansion" ], "citationOrigin": "explicit-claim-map", "evidenceMode": "editorial-analysis", "evidenceNote": "Briefing field inherits the entry source.", "numericValues": [], "isNumeric": false, "citationCount": 1, "validSourceIds": [ "tsmc-arizona-phase23-expansion" ], "invalidSourceIds": [], "sourceDetails": [ { "id": "tsmc-arizona-phase23-expansion", "title": "TSMC Board Authorization of $100B Multi-Phase Phoenix Arizona Foundry Expansion", "publisher": "TSMC Investor Relations / SEC Form 6-K", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true } ], "bestGrade": "A", "worstGrade": "A", "coverage": { "status": "supported", "reason": "At least one valid source explicitly supports this claim." } }, { "id": "claim:executiveBriefing.2.summary", "path": "executiveBriefing.2.summary", "section": "executive-briefing", "kind": "synthesis", "critical": true, "text": "Anthropic mandates POSIX microVM sandboxing for Claude Motion desktop agents.", "sourceIds": [ "anthropic-claude-motion-posix-policy" ], "citationOrigin": "explicit-claim-map", "evidenceMode": "editorial-analysis", "evidenceNote": "Briefing field inherits the entry source.", "numericValues": [], "isNumeric": false, "citationCount": 1, "validSourceIds": [ "anthropic-claude-motion-posix-policy" ], "invalidSourceIds": [], "sourceDetails": [ { "id": "anthropic-claude-motion-posix-policy", "title": "Anthropic Claude Motion Engine GA and Mandatory Host POSIX MicroVM Containment Policy", "publisher": "Anthropic PBC", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true } ], "bestGrade": "A", "worstGrade": "A", "coverage": { "status": "supported", "reason": "At least one valid source explicitly supports this claim." } }, { "id": "claim:executiveBriefing.2.whatHappened", "path": "executiveBriefing.2.whatHappened", "section": "executive-briefing", "kind": "factual", "critical": true, "text": "Anthropic launched Claude Motion with Firecracker microVM containment and host eBPF system call filtering.", "sourceIds": [ "anthropic-claude-motion-posix-policy" ], "citationOrigin": "explicit-claim-map", "evidenceMode": "editorial-analysis", "evidenceNote": "Briefing field inherits the entry source.", "numericValues": [], "isNumeric": false, "citationCount": 1, "validSourceIds": [ "anthropic-claude-motion-posix-policy" ], "invalidSourceIds": [], "sourceDetails": [ { "id": "anthropic-claude-motion-posix-policy", "title": "Anthropic Claude Motion Engine GA and Mandatory Host POSIX MicroVM Containment Policy", "publisher": "Anthropic PBC", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true } ], "bestGrade": "A", "worstGrade": "A", "coverage": { "status": "supported", "reason": "At least one valid source explicitly supports this claim." } }, { "id": "claim:executiveBriefing.2.whyItMatters", "path": "executiveBriefing.2.whyItMatters", "section": "executive-briefing", "kind": "analysis", "critical": true, "text": "Establishes hardware-assisted containerization as the non-negotiable security baseline for autonomous computer-use agents.", "sourceIds": [ "anthropic-claude-motion-posix-policy" ], "citationOrigin": "explicit-claim-map", "evidenceMode": "editorial-analysis", "evidenceNote": "Briefing field inherits the entry source.", "numericValues": [], "isNumeric": false, "citationCount": 1, "validSourceIds": [ "anthropic-claude-motion-posix-policy" ], "invalidSourceIds": [], "sourceDetails": [ { "id": "anthropic-claude-motion-posix-policy", "title": "Anthropic Claude Motion Engine GA and Mandatory Host POSIX MicroVM Containment Policy", "publisher": "Anthropic PBC", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true } ], "bestGrade": "A", "worstGrade": "A", "coverage": { "status": "supported", "reason": "At least one valid source explicitly supports this claim." } }, { "id": "claim:executiveBriefing.2.action", "path": "executiveBriefing.2.action", "section": "executive-briefing", "kind": "recommendation", "critical": false, "text": "Upgrade autonomous agent tool environments from shared Docker daemons to ephemeral microVM sandboxes.", "sourceIds": [ "anthropic-claude-motion-posix-policy" ], "citationOrigin": "explicit-claim-map", "evidenceMode": "editorial-analysis", "evidenceNote": "Briefing field inherits the entry source.", "numericValues": [], "isNumeric": false, "citationCount": 1, "validSourceIds": [ "anthropic-claude-motion-posix-policy" ], "invalidSourceIds": [], "sourceDetails": [ { "id": "anthropic-claude-motion-posix-policy", "title": "Anthropic Claude Motion Engine GA and Mandatory Host POSIX MicroVM Containment Policy", "publisher": "Anthropic PBC", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true } ], "bestGrade": "A", "worstGrade": "A", "coverage": { "status": "supported", "reason": "At least one valid source explicitly supports this claim." } }, { "id": "claim:snapshots.0", "path": "snapshots.0", "section": "snapshot-cards", "kind": "structured-summary", "critical": true, "text": "SWE-bench Verified Leaderboard: 78.4%. +8.1% vs Claude 3.7. undefined", "sourceIds": [ "openai-o3-pro-system-card" ], "citationOrigin": "record-level", "evidenceMode": "direct", "evidenceNote": null, "numericValues": [ "78.4%", "8.1%" ], "isNumeric": true, "citationCount": 1, "validSourceIds": [ "openai-o3-pro-system-card" ], "invalidSourceIds": [], "sourceDetails": [ { "id": "openai-o3-pro-system-card", "title": "OpenAI o3-pro System Card: Frontier Reasoning, Test-Time Tree Search with Native Multi-Step Tool Execution, and Autonomous Coding Evaluations", "publisher": "OpenAI", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true } ], "bestGrade": "A", "worstGrade": "A", "coverage": { "status": "supported", "reason": "At least one valid source explicitly supports this claim." } }, { "id": "claim:snapshots.1", "path": "snapshots.1", "section": "snapshot-cards", "kind": "structured-summary", "critical": true, "text": "2nm N2P Power Delivery: -85% IR Drop. Super Power Rail Tape-Out. undefined", "sourceIds": [ "tsmc-n2p-gaa-bspdn-announcement" ], "citationOrigin": "record-level", "evidenceMode": "direct", "evidenceNote": null, "numericValues": [ "85%" ], "isNumeric": true, "citationCount": 1, "validSourceIds": [ "tsmc-n2p-gaa-bspdn-announcement" ], "invalidSourceIds": [], "sourceDetails": [ { "id": "tsmc-n2p-gaa-bspdn-announcement", "title": "TSMC Completes Tape-Out of 2nm (N2P) GAA Silicon with Backside Power Delivery (Super Power Rail) for Next-Generation AI Accelerators", "publisher": "TSMC / NVIDIA Press Center", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true } ], "bestGrade": "A", "worstGrade": "A", "coverage": { "status": "supported", "reason": "At least one valid source explicitly supports this claim." } }, { "id": "claim:snapshots.2", "path": "snapshots.2", "section": "snapshot-cards", "kind": "structured-summary", "critical": true, "text": "Open-Weight Coding Efficiency: 65.8% SWE-bench. Apache 2.0 Open Weights. undefined", "sourceIds": [ "alibaba-qwen-2-5-coder-reasoner-paper" ], "citationOrigin": "record-level", "evidenceMode": "direct", "evidenceNote": null, "numericValues": [ "65.8%" ], "isNumeric": true, "citationCount": 1, "validSourceIds": [ "alibaba-qwen-2-5-coder-reasoner-paper" ], "invalidSourceIds": [], "sourceDetails": [ { "id": "alibaba-qwen-2-5-coder-reasoner-paper", "title": "Qwen 2.5-Coder-32B-Reasoner: Open-Weight Test-Time Search Code Agent with Step-Level Process Reward Supervision", "publisher": "Alibaba Cloud & Qwen Team", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true } ], "bestGrade": "A", "worstGrade": "A", "coverage": { "status": "supported", "reason": "At least one valid source explicitly supports this claim." } }, { "id": "claim:snapshots.3", "path": "snapshots.3", "section": "snapshot-cards", "kind": "structured-summary", "critical": true, "text": "Agent Steganography Bypasses: 88% Bypass Rate. MicroVM Sandbox Mandated. undefined", "sourceIds": [ "aisi-joint-multimodal-agent-sandbox-audit" ], "citationOrigin": "record-level", "evidenceMode": "direct", "evidenceNote": null, "numericValues": [ "88%" ], "isNumeric": true, "citationCount": 1, "validSourceIds": [ "aisi-joint-multimodal-agent-sandbox-audit" ], "invalidSourceIds": [], "sourceDetails": [ { "id": "aisi-joint-multimodal-agent-sandbox-audit", "title": "US and UK AI Safety Institutes: Joint Empirical Audit on Multimodal Visual Prompt Smuggling and Mandatory Sandbox Virtualization for Desktop Agents", "publisher": "US AI Safety Institute (NIST) / UK AI Safety Institute (DSIT)", "grade": "C", "score": 71, "freshnessStatus": "current", "critical": true } ], "bestGrade": "C", "worstGrade": "C", "coverage": { "status": "supported", "reason": "At least one valid source explicitly supports this claim." } }, { "id": "claim:models.0", "path": "models.0", "section": "models", "kind": "structured-record", "critical": true, "text": "Google DeepMind Gemini 3.8 (Flash & Pro) — released: 2026-10-15; availability: Google AI Studio, Vertex AI, Gemini Advanced; context: 2,000,000 native; modalities: Text, Vision, Audio, Video, Code, Tool Use; strength: Native 2M context with sub-125ms live multimodal streaming, top-ranked general reasoning (65.4 on General365), and integrated SynthID Universal provenance watermarks; input price: 0.075; output price: 0.3; cached input price: 0.019; price currency: USD", "sourceIds": [ "google-gemini-38-multimodal-release" ], "citationOrigin": "record-level", "evidenceMode": "direct", "evidenceNote": null, "numericValues": [ "0.075", "0.3", "0.019", "2M" ], "isNumeric": true, "citationCount": 1, "validSourceIds": [ "google-gemini-38-multimodal-release" ], "invalidSourceIds": [], "sourceDetails": [ { "id": "google-gemini-38-multimodal-release", "title": "Google DeepMind Gemini 3.8 (Flash & Pro) Architecture Whitepaper & System API Documentation", "publisher": "Google DeepMind", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true } ], "bestGrade": "A", "worstGrade": "A", "coverage": { "status": "supported", "reason": "At least one valid source explicitly supports this claim." } }, { "id": "claim:models.1", "path": "models.1", "section": "models", "kind": "structured-record", "critical": true, "text": "OpenAI GPT-6 (Sol & Luna) — released: 2026-10-14; availability: ChatGPT Enterprise, Pro, Plus, API; context: 256,000 native; modalities: Text, Vision, Audio, UI Components, Canvas Tools; strength: Dual-engine speculative decoding runtime (Sol 180 tok/s, Luna deep reasoning verifier) with native declarative Intelligent UI component trees; input price: 0.35; output price: 1.4; cached input price: 0.08; price currency: USD", "sourceIds": [ "openai-gpt6-ga-whitepaper" ], "citationOrigin": "record-level", "evidenceMode": "direct", "evidenceNote": null, "numericValues": [ "0.35", "1.4", "0.08" ], "isNumeric": true, "citationCount": 1, "validSourceIds": [ "openai-gpt6-ga-whitepaper" ], "invalidSourceIds": [], "sourceDetails": [ { "id": "openai-gpt6-ga-whitepaper", "title": "OpenAI GPT-6 (Sol & Luna) Dual-Engine Production Deployment Specification", "publisher": "OpenAI", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true } ], "bestGrade": "A", "worstGrade": "A", "coverage": { "status": "supported", "reason": "At least one valid source explicitly supports this claim." } }, { "id": "claim:models.2", "path": "models.2", "section": "models", "kind": "structured-record", "critical": true, "text": "Anthropic Claude 3.8 (Workspaces & Motion) — released: 2026-10-16; availability: Anthropic API, Claude Enterprise, Claude Pro; context: 200,000 native; modalities: Text, Vision, Code, Filesystem Actions, WebGL Canvas; strength: Native host POSIX microVM sandboxing, collaborative workspace artifacts GA, and Motion dynamic animated WebGL visualization engine; input price: 3; output price: 15; cached input price: 0.75; price currency: USD", "sourceIds": [ "anthropic-claude-motion-posix-policy" ], "citationOrigin": "record-level", "evidenceMode": "direct", "evidenceNote": null, "numericValues": [ "3", "15", "0.75" ], "isNumeric": true, "citationCount": 1, "validSourceIds": [ "anthropic-claude-motion-posix-policy" ], "invalidSourceIds": [], "sourceDetails": [ { "id": "anthropic-claude-motion-posix-policy", "title": "Anthropic Claude Motion Engine GA and Mandatory Host POSIX MicroVM Containment Policy", "publisher": "Anthropic PBC", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true } ], "bestGrade": "A", "worstGrade": "A", "coverage": { "status": "supported", "reason": "At least one valid source explicitly supports this claim." } }, { "id": "claim:models.3", "path": "models.3", "section": "models", "kind": "structured-record", "critical": true, "text": "Alibaba Qwen-Image-2.1-Turbo — released: 2026-10-15; availability: Open Weights (Hugging Face / ModelScope), Alibaba Cloud API; context: 8,192 text & reference tokens; modalities: Text-to-Image, Multi-Reference Editing, Transparent RGBA; strength: Accelerated 8-step Flow Matching Euler diffusion with 64-channel RGBA VAE for native 2K transparent image synthesis (418 ms on H100); input price: 0.014; output price: 0.014; cached input price: 0.007; price currency: USD", "sourceIds": [ "openai-gpt6-ga-whitepaper" ], "citationOrigin": "record-level", "evidenceMode": "direct", "evidenceNote": null, "numericValues": [ "0.014", "0.007", "2K" ], "isNumeric": true, "citationCount": 1, "validSourceIds": [ "openai-gpt6-ga-whitepaper" ], "invalidSourceIds": [], "sourceDetails": [ { "id": "openai-gpt6-ga-whitepaper", "title": "OpenAI GPT-6 (Sol & Luna) Dual-Engine Production Deployment Specification", "publisher": "OpenAI", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true } ], "bestGrade": "A", "worstGrade": "A", "coverage": { "status": "supported", "reason": "At least one valid source explicitly supports this claim." } }, { "id": "claim:models.4", "path": "models.4", "section": "models", "kind": "structured-record", "critical": true, "text": "DeepSeek DeepSeek-V3.5 — released: 2026-10-09; availability: Open Weights (Hugging Face / GitHub), DeepSeek API, Huawei Ascend CANN; context: 256,000 native; modalities: Text, Code; strength: Multi-Head Latent Attention (75% KV-cache reduction) and DualPipe scheduling optimized natively across 160,000 Huawei Ascend 910C nodes; input price: 0.14; output price: 0.28; cached input price: 0.028; price currency: USD", "sourceIds": [ "deepseek-pre-ipo-ascend-cluster" ], "citationOrigin": "record-level", "evidenceMode": "direct", "evidenceNote": null, "numericValues": [ "0.14", "0.28", "0.028", "75%" ], "isNumeric": true, "citationCount": 1, "validSourceIds": [ "deepseek-pre-ipo-ascend-cluster" ], "invalidSourceIds": [], "sourceDetails": [ { "id": "deepseek-pre-ipo-ascend-cluster", "title": "DeepSeek Infrastructure Architecture: 160,000-Accelerator Huawei Ascend CANN Cluster & Capital Disclosures", "publisher": "DeepSeek AI / Industry Regulatory Filings", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true } ], "bestGrade": "A", "worstGrade": "A", "coverage": { "status": "supported", "reason": "At least one valid source explicitly supports this claim." } }, { "id": "claim:models.5", "path": "models.5", "section": "models", "kind": "structured-record", "critical": true, "text": "OpenAI OpenAI Operator (Web Agent) — released: 2026-10-07; availability: ChatGPT Pro, ChatGPT Team, OpenAI API (Pilot); context: 128,000 vision & DOM context; modalities: Text, Vision, Web Browser, Native DOM Tools; strength: Autonomous multi-tab browser execution inside ephemeral MicroVM sandboxes (WebArena 68.2%, OSWorld 44.5%), real-time DOM action verification; input price: 5; output price: 20; cached input price: 1.25; price currency: USD", "sourceIds": [ "openai-operator-sandbox-architecture" ], "citationOrigin": "record-level", "evidenceMode": "direct", "evidenceNote": null, "numericValues": [ "5", "20", "1.25", "68.2%", "44.5%" ], "isNumeric": true, "citationCount": 1, "validSourceIds": [ "openai-operator-sandbox-architecture" ], "invalidSourceIds": [], "sourceDetails": [ { "id": "openai-operator-sandbox-architecture", "title": "OpenAI Operator Architecture: MicroVM Hardware Isolation, Permission Gating, and Dual-Loop Verification for Web Agents", "publisher": "OpenAI", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true } ], "bestGrade": "A", "worstGrade": "A", "coverage": { "status": "supported", "reason": "At least one valid source explicitly supports this claim." } }, { "id": "claim:models.6", "path": "models.6", "section": "models", "kind": "structured-record", "critical": true, "text": "Google Gemini 2.5 Pro (Thinking Mode) — released: 2026-10-06; availability: Google Cloud Vertex AI, Google AI Studio, Gemini Advanced; context: 1,000,000 native (up to 2M in Vertex AI); modalities: Text, Code, Vision, Audio, Video; strength: Native recursive scratchpad execution with sandboxed Python compiler tool calling; excels at massive document analysis and multi-step math reasoning; input price: 1.25; output price: 5; cached input price: 0.31; price currency: USD", "sourceIds": [ "google-deepmind-gemini-2-5-thinking-vertex" ], "citationOrigin": "record-level", "evidenceMode": "direct", "evidenceNote": null, "numericValues": [ "2M", "1.25", "5", "0.31" ], "isNumeric": true, "citationCount": 1, "validSourceIds": [ "google-deepmind-gemini-2-5-thinking-vertex" ], "invalidSourceIds": [], "sourceDetails": [ { "id": "google-deepmind-gemini-2-5-thinking-vertex", "title": "Google Cloud Vertex AI: Gemini 2.5 Flash and Pro Thinking Mode with Native Recursive Code Execution", "publisher": "Google Cloud / Google DeepMind", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true } ], "bestGrade": "A", "worstGrade": "A", "coverage": { "status": "supported", "reason": "At least one valid source explicitly supports this claim." } }, { "id": "claim:models.7", "path": "models.7", "section": "models", "kind": "structured-record", "critical": true, "text": "Meta Llama 4 Scout (Architecture Preview) — released: 2026-10-07; availability: Hugging Face, Meta AI Research, GitHub; context: 256,000 native; modalities: Text, Code; strength: Sparse Mixture-of-Experts with native draft-model speculative decoding, delivering 2.8x higher throughput on 8x H100 clusters (MATH-500 89.4%, GSM8K 96.2%); input price: 0.2; output price: 0.6; cached input price: 0.05; price currency: USD", "sourceIds": [ "meta-llama-4-speculative-decoding-preview" ], "citationOrigin": "record-level", "evidenceMode": "direct", "evidenceNote": null, "numericValues": [ "0.2", "0.6", "0.05", "89.4%", "96.2%" ], "isNumeric": true, "citationCount": 1, "validSourceIds": [ "meta-llama-4-speculative-decoding-preview" ], "invalidSourceIds": [], "sourceDetails": [ { "id": "meta-llama-4-speculative-decoding-preview", "title": "Meta AI Engineering: Llama 4 Architectural Blueprint, Native Speculative Decoding, and MoE Routing at Scale", "publisher": "Meta AI / Fundamental AI Research", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true } ], "bestGrade": "A", "worstGrade": "A", "coverage": { "status": "supported", "reason": "At least one valid source explicitly supports this claim." } }, { "id": "claim:models.8", "path": "models.8", "section": "models", "kind": "structured-record", "critical": true, "text": "Anthropic Claude 3.7 Sonnet — released: 2026-10-05; availability: Claude.ai (Pro/Team/Free), Anthropic API, Amazon Bedrock, Google Cloud Vertex AI; context: 200k native; modalities: Text, Code, Vision, Native GUI Tools; strength: Dual-mode hybrid reasoning with user-defined thinking budgets (SWE-bench Verified 70.3%, Terminal-Bench 58.2%, GPQA Diamond 65.1%) and 60 FPS Computer Use v2; input price: 3; output price: 15; cached input price: 0.3; price currency: USD", "sourceIds": [ "anthropic-claude-3-7-sonnet-system-card" ], "citationOrigin": "record-level", "evidenceMode": "direct", "evidenceNote": null, "numericValues": [ "200k", "3", "15", "0.3", "70.3%", "58.2%", "65.1%" ], "isNumeric": true, "citationCount": 1, "validSourceIds": [ "anthropic-claude-3-7-sonnet-system-card" ], "invalidSourceIds": [], "sourceDetails": [ { "id": "anthropic-claude-3-7-sonnet-system-card", "title": "Anthropic Claude 3.7 Sonnet: Hybrid Reasoning Architecture, Extended Thinking Token Scaling, and Sub-Pixel Computer Use v2", "publisher": "Anthropic", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true } ], "bestGrade": "A", "worstGrade": "A", "coverage": { "status": "supported", "reason": "At least one valid source explicitly supports this claim." } }, { "id": "claim:models.9", "path": "models.9", "section": "models", "kind": "structured-record", "critical": true, "text": "OpenAI OpenAI o3-pro — released: 2026-09-28; availability: OpenAI API, ChatGPT Pro, Microsoft Azure OpenAI Service; context: 200k native (100k max completion); modalities: Text, Code, Vision, Native Tools; strength: Deep test-time tree search with integrated tool calling (SWE-bench Verified 78.4%, MATH-500 92.1%, FrontierMath 34.2%, GPQA Diamond 87.6%); input price: 15; output price: 60; cached input price: 3.75; price currency: USD", "sourceIds": [ "openai-o3-pro-system-card" ], "citationOrigin": "record-level", "evidenceMode": "direct", "evidenceNote": null, "numericValues": [ "200k", "100k", "15", "60", "3.75", "78.4%", "92.1%", "34.2%", "87.6%" ], "isNumeric": true, "citationCount": 1, "validSourceIds": [ "openai-o3-pro-system-card" ], "invalidSourceIds": [], "sourceDetails": [ { "id": "openai-o3-pro-system-card", "title": "OpenAI o3-pro System Card: Frontier Reasoning, Test-Time Tree Search with Native Multi-Step Tool Execution, and Autonomous Coding Evaluations", "publisher": "OpenAI", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true } ], "bestGrade": "A", "worstGrade": "A", "coverage": { "status": "supported", "reason": "At least one valid source explicitly supports this claim." } }, { "id": "claim:models.10", "path": "models.10", "section": "models", "kind": "structured-record", "critical": true, "text": "Alibaba Cloud Qwen 2.5-Coder-32B-Reasoner — released: 2026-09-27; availability: Hugging Face, ModelScope, Alibaba Cloud DashScope, vLLM; context: 128k native; modalities: Text, Code; strength: Open-weights (Apache 2.0), step-level Process Reward Model scoring, SWE-bench Verified 65.8%, HumanEval+ 81.2%, runs on dual RTX 4090 GPUs in FP8; input price: 0.14; output price: 0.42; cached input price: 0.035; price currency: USD", "sourceIds": [ "alibaba-qwen-2-5-coder-reasoner-paper" ], "citationOrigin": "record-level", "evidenceMode": "direct", "evidenceNote": null, "numericValues": [ "128k", "0.14", "0.42", "0.035", "32B", "65.8%", "81.2%" ], "isNumeric": true, "citationCount": 1, "validSourceIds": [ "alibaba-qwen-2-5-coder-reasoner-paper" ], "invalidSourceIds": [], "sourceDetails": [ { "id": "alibaba-qwen-2-5-coder-reasoner-paper", "title": "Qwen 2.5-Coder-32B-Reasoner: Open-Weight Test-Time Search Code Agent with Step-Level Process Reward Supervision", "publisher": "Alibaba Cloud & Qwen Team", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true } ], "bestGrade": "A", "worstGrade": "A", "coverage": { "status": "supported", "reason": "At least one valid source explicitly supports this claim." } }, { "id": "claim:models.11", "path": "models.11", "section": "models", "kind": "structured-record", "critical": true, "text": "Anthropic Claude 3.7 Sonnet (Hybrid Thinking Mode) — released: 2026-09-24; availability: Anthropic API, Claude Enterprise, AWS Bedrock, Google Cloud Vertex AI; context: 200k native (128k max completion); modalities: Text, Code, Vision; strength: Dynamic thinking budgets (1k-64k tokens), SWE-bench Verified 70.3%, HumanEval 96.2%, GPQA Diamond 84.8%, Computer Use 2.0 with 40% lower action latency; input price: 3; output price: 15; cached input price: 0.3; price currency: USD", "sourceIds": [ "anthropic-claude-3-7-sonnet-system-card" ], "citationOrigin": "record-level", "evidenceMode": "direct", "evidenceNote": null, "numericValues": [ "200k", "128k", "3", "15", "0.3", "1k", "64k", "70.3%", "96.2%", "84.8%", "40%" ], "isNumeric": true, "citationCount": 1, "validSourceIds": [ "anthropic-claude-3-7-sonnet-system-card" ], "invalidSourceIds": [], "sourceDetails": [ { "id": "anthropic-claude-3-7-sonnet-system-card", "title": "Anthropic Claude 3.7 Sonnet: Hybrid Reasoning Architecture, Extended Thinking Token Scaling, and Sub-Pixel Computer Use v2", "publisher": "Anthropic", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true } ], "bestGrade": "A", "worstGrade": "A", "coverage": { "status": "supported", "reason": "At least one valid source explicitly supports this claim." } }, { "id": "claim:models.12", "path": "models.12", "section": "models", "kind": "structured-record", "critical": true, "text": "Anthropic Claude 3.7 Sonnet (Standard Latency Mode) — released: 2026-09-24; availability: Anthropic API, Claude Free/Pro/Team/Enterprise; context: 200k native (128k max completion); modalities: Text, Code, Vision; strength: Instant-generation flagship performance without scratchpad latency, state-of-the-art general instruction following and natural voice prose; input price: 3; output price: 15; cached input price: 0.3; price currency: USD", "sourceIds": [ "anthropic-claude-3-7-sonnet-system-card" ], "citationOrigin": "record-level", "evidenceMode": "direct", "evidenceNote": null, "numericValues": [ "200k", "128k", "3", "15", "0.3" ], "isNumeric": true, "citationCount": 1, "validSourceIds": [ "anthropic-claude-3-7-sonnet-system-card" ], "invalidSourceIds": [], "sourceDetails": [ { "id": "anthropic-claude-3-7-sonnet-system-card", "title": "Anthropic Claude 3.7 Sonnet: Hybrid Reasoning Architecture, Extended Thinking Token Scaling, and Sub-Pixel Computer Use v2", "publisher": "Anthropic", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true } ], "bestGrade": "A", "worstGrade": "A", "coverage": { "status": "supported", "reason": "At least one valid source explicitly supports this claim." } }, { "id": "claim:models.13", "path": "models.13", "section": "models", "kind": "structured-record", "critical": true, "text": "OpenAI o3-mini (High Effort) — released: 2026-09-22; availability: OpenAI Developer API, ChatGPT Plus/Team/Enterprise; context: 200k native (100k max completion/reasoning); modalities: Text, Code; strength: Controllable test-time search (MATH-500 88.2%, SWE-bench Verified 59.4%, AIME 2025 79.2%), eBPF-isolated microVM execution; input price: 0.4; output price: 1.6; cached input price: 0.2; price currency: USD", "sourceIds": [ "openai-o3-mini-system-card" ], "citationOrigin": "record-level", "evidenceMode": "direct", "evidenceNote": null, "numericValues": [ "200k", "100k", "0.4", "1.6", "0.2", "88.2%", "59.4%", "79.2%" ], "isNumeric": true, "citationCount": 1, "validSourceIds": [ "openai-o3-mini-system-card" ], "invalidSourceIds": [], "sourceDetails": [ { "id": "openai-o3-mini-system-card", "title": "OpenAI o3-mini System Card & Technical Documentation: Controllable Test-Time Compute, Mathematical Reasoning, and MicroVM Sandboxing", "publisher": "OpenAI", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true } ], "bestGrade": "A", "worstGrade": "A", "coverage": { "status": "supported", "reason": "At least one valid source explicitly supports this claim." } }, { "id": "claim:models.14", "path": "models.14", "section": "models", "kind": "structured-record", "critical": true, "text": "OpenAI o3-mini (Medium Effort) — released: 2026-09-22; availability: OpenAI Developer API, ChatGPT Plus/Team/Enterprise; context: 200k native; modalities: Text, Code; strength: Balanced latency/reasoning trade-off (MATH-500 84.8%, SWE-bench Verified 51.2%), fast structured tool calling; input price: 0.4; output price: 1.6; cached input price: 0.2; price currency: USD", "sourceIds": [ "openai-o3-mini-system-card" ], "citationOrigin": "record-level", "evidenceMode": "direct", "evidenceNote": null, "numericValues": [ "200k", "0.4", "1.6", "0.2", "84.8%", "51.2%" ], "isNumeric": true, "citationCount": 1, "validSourceIds": [ "openai-o3-mini-system-card" ], "invalidSourceIds": [], "sourceDetails": [ { "id": "openai-o3-mini-system-card", "title": "OpenAI o3-mini System Card & Technical Documentation: Controllable Test-Time Compute, Mathematical Reasoning, and MicroVM Sandboxing", "publisher": "OpenAI", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true } ], "bestGrade": "A", "worstGrade": "A", "coverage": { "status": "supported", "reason": "At least one valid source explicitly supports this claim." } }, { "id": "claim:models.15", "path": "models.15", "section": "models", "kind": "structured-record", "critical": true, "text": "Google Gemini 2.5 Pro Reasoner — released: 2026-09-21; availability: Google AI Studio, Vertex AI, Gemini API; context: 1M native; modalities: Text, Code, Audio, Video, ROS2 Actions; strength: Test-time tree search (MATH-500 86.4%, SWE-bench Verified 61.2%, FrontierMath 28.4%), integrated robotics spatial grounding; input price: 1.25; output price: 5; cached input price: 0.3125; price currency: USD", "sourceIds": [ "google-gemini-2-5-pro-reasoner-paper" ], "citationOrigin": "record-level", "evidenceMode": "direct", "evidenceNote": null, "numericValues": [ "1M", "1.25", "5", "0.3125", "86.4%", "61.2%", "28.4%" ], "isNumeric": true, "citationCount": 1, "validSourceIds": [ "google-gemini-2-5-pro-reasoner-paper" ], "invalidSourceIds": [], "sourceDetails": [ { "id": "google-gemini-2-5-pro-reasoner-paper", "title": "Gemini 2.5 Pro Reasoner Technical Report: Step-Level Supervision, Monte Carlo Tree Search, and Robotics Foundation API", "publisher": "Google DeepMind", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true } ], "bestGrade": "A", "worstGrade": "A", "coverage": { "status": "supported", "reason": "At least one valid source explicitly supports this claim." } }, { "id": "claim:models.16", "path": "models.16", "section": "models", "kind": "structured-record", "critical": true, "text": "Mistral AI Mistral Large 3 — released: 2026-09-21; availability: Open weights (Mistral Research License), Mistral La Plateforme; context: 1M native; modalities: Text, Code; strength: Hybrid 3:1 Mamba-3 SSM / Transformer architecture, 78.4% reduction in KV-cache VRAM footprint, RULER 256k 88.6%; input price: 0.2; output price: 0.6; cached input price: 0.05; price currency: USD", "sourceIds": [ "mistral-large-3-hybrid-ssm-paper" ], "citationOrigin": "record-level", "evidenceMode": "direct", "evidenceNote": null, "numericValues": [ "1M", "0.2", "0.6", "0.05", "78.4%", "256k", "88.6%" ], "isNumeric": true, "citationCount": 1, "validSourceIds": [ "mistral-large-3-hybrid-ssm-paper" ], "invalidSourceIds": [], "sourceDetails": [ { "id": "mistral-large-3-hybrid-ssm-paper", "title": "Mistral Large 3: Dense 145B Hybrid Selective State-Space and Transformer Model Architecture", "publisher": "Mistral AI Research", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true } ], "bestGrade": "A", "worstGrade": "A", "coverage": { "status": "supported", "reason": "At least one valid source explicitly supports this claim." } }, { "id": "claim:models.17", "path": "models.17", "section": "models", "kind": "structured-record", "critical": true, "text": "Meta Llama 4 Frontier Preview — released: 2026-09-17; availability: Open weights (research license), Hugging Face, Meta AI; context: 256k native; modalities: Text, Code; strength: Sparse MoE reasoning (MMLU-Pro 78.4%), automated software engineering (SWE-bench Verified 52.6%), sub-byte serving efficiency; input price: 0.18; output price: 0.54; cached input price: 0.045; price currency: USD", "sourceIds": [ "meta-llama4-frontier-preview-spec" ], "citationOrigin": "record-level", "evidenceMode": "direct", "evidenceNote": null, "numericValues": [ "256k", "0.18", "0.54", "0.045", "78.4%", "52.6%" ], "isNumeric": true, "citationCount": 1, "validSourceIds": [ "meta-llama4-frontier-preview-spec" ], "invalidSourceIds": [], "sourceDetails": [ { "id": "meta-llama4-frontier-preview-spec", "title": "Llama 4 Frontier Preview Architecture Technical Report: 128-Expert Sparse MoE and Native MXFP4 Training at Scale", "publisher": "Meta AI Research", "grade": "A", "score": 93, "freshnessStatus": "stale", "critical": true } ], "bestGrade": "A", "worstGrade": "A", "coverage": { "status": "supported", "reason": "At least one valid source explicitly supports this claim." } }, { "id": "claim:models.18", "path": "models.18", "section": "models", "kind": "structured-record", "critical": true, "text": "DeepSeek V4 Pro — released: 2026-08-13; availability: DeepSeek API, app and web; context: 1M; modalities: Text; strength: Long-context reasoning, coding and agent workflows; input price: 0.435; output price: 0.87; cached input price: 0.003625; price currency: USD", "sourceIds": [ "reuters-deepseek-v4-0813-aug13" ], "citationOrigin": "record-level", "evidenceMode": "direct", "evidenceNote": null, "numericValues": [ "1M", "0.435", "0.87", "0.003625" ], "isNumeric": true, "citationCount": 1, "validSourceIds": [ "reuters-deepseek-v4-0813-aug13" ], "invalidSourceIds": [], "sourceDetails": [ { "id": "reuters-deepseek-v4-0813-aug13", "title": "DeepSeek releases official V4 Pro model as it steps up expansion", "publisher": "Reuters", "grade": "D", "score": 55, "freshnessStatus": "expired", "critical": true } ], "bestGrade": "D", "worstGrade": "D", "coverage": { "status": "weak", "reason": "Every attached source is below the preferred evidence-quality threshold." } }, { "id": "claim:models.19", "path": "models.19", "section": "models", "kind": "structured-record", "critical": true, "text": "Anthropic Claude Opus 5 — released: 2026-07-24; availability: Anthropic API, Amazon Bedrock, Google Cloud and Microsoft Foundry; context: 1M; modalities: Text and image input; text output; strength: Long-horizon reasoning, interactive agent work and complex professional tasks; input price: 5; output price: 25; price currency: USD", "sourceIds": [ "anthropic-models-overview-aug9" ], "citationOrigin": "record-level", "evidenceMode": "direct", "evidenceNote": null, "numericValues": [ "1M", "5", "25" ], "isNumeric": true, "citationCount": 1, "validSourceIds": [ "anthropic-models-overview-aug9" ], "invalidSourceIds": [], "sourceDetails": [ { "id": "anthropic-models-overview-aug9", "title": "Claude models overview", "publisher": "Anthropic", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true } ], "bestGrade": "B", "worstGrade": "B", "coverage": { "status": "supported", "reason": "At least one valid source explicitly supports this claim." } }, { "id": "claim:models.20", "path": "models.20", "section": "models", "kind": "structured-record", "critical": true, "text": "OpenAI GPT-5.6 Sol — released: 2026-07-09; availability: API and eligible ChatGPT plans; context: See current model documentation; modalities: Text, vision and tool-enabled workflows; strength: Premium reasoning and agentic work; input price: 5; output price: 30; price currency: USD", "sourceIds": [ "openai-gpt56" ], "citationOrigin": "record-level", "evidenceMode": "direct", "evidenceNote": null, "numericValues": [ "5", "30" ], "isNumeric": true, "citationCount": 1, "validSourceIds": [ "openai-gpt56" ], "invalidSourceIds": [], "sourceDetails": [ { "id": "openai-gpt56", "title": "Introducing GPT-5.6", "publisher": "OpenAI", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true } ], "bestGrade": "B", "worstGrade": "B", "coverage": { "status": "supported", "reason": "At least one valid source explicitly supports this claim." } }, { "id": "claim:models.21", "path": "models.21", "section": "models", "kind": "structured-record", "critical": true, "text": "OpenAI GPT-5.6 Terra — released: 2026-07-09; availability: API and eligible ChatGPT plans; context: See current model documentation; modalities: Text, vision and tool-enabled workflows; strength: Mid-tier capability and cost balance; input price: 2.5; output price: 15; price currency: USD", "sourceIds": [ "openai-gpt56" ], "citationOrigin": "record-level", "evidenceMode": "direct", "evidenceNote": null, "numericValues": [ "2.5", "15" ], "isNumeric": true, "citationCount": 1, "validSourceIds": [ "openai-gpt56" ], "invalidSourceIds": [], "sourceDetails": [ { "id": "openai-gpt56", "title": "Introducing GPT-5.6", "publisher": "OpenAI", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true } ], "bestGrade": "B", "worstGrade": "B", "coverage": { "status": "supported", "reason": "At least one valid source explicitly supports this claim." } }, { "id": "claim:models.22", "path": "models.22", "section": "models", "kind": "structured-record", "critical": true, "text": "OpenAI GPT-5.6 Luna — released: 2026-07-09; availability: API and eligible ChatGPT plans; context: See current model documentation; modalities: Text and general assistant workflows; strength: Lowest-cost member of the GPT-5.6 family; input price: 1; output price: 6; price currency: USD", "sourceIds": [ "openai-gpt56" ], "citationOrigin": "record-level", "evidenceMode": "direct", "evidenceNote": null, "numericValues": [ "1", "6" ], "isNumeric": true, "citationCount": 1, "validSourceIds": [ "openai-gpt56" ], "invalidSourceIds": [], "sourceDetails": [ { "id": "openai-gpt56", "title": "Introducing GPT-5.6", "publisher": "OpenAI", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true } ], "bestGrade": "B", "worstGrade": "B", "coverage": { "status": "supported", "reason": "At least one valid source explicitly supports this claim." } }, { "id": "claim:models.23", "path": "models.23", "section": "models", "kind": "structured-record", "critical": true, "text": "Anthropic Claude Fable 5 — released: 2026-06; availability: Claude and API where available; context: See current model documentation; modalities: Text, vision, coding and long-running agents; strength: Premium long-running agent work; input price: 10; output price: 50; price currency: USD", "sourceIds": [ "anthropic-fable5" ], "citationOrigin": "record-level", "evidenceMode": "direct", "evidenceNote": null, "numericValues": [ "10", "50" ], "isNumeric": true, "citationCount": 1, "validSourceIds": [ "anthropic-fable5" ], "invalidSourceIds": [], "sourceDetails": [ { "id": "anthropic-fable5", "title": "Claude Fable 5 and Mythos 5", "publisher": "Anthropic", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true } ], "bestGrade": "B", "worstGrade": "B", "coverage": { "status": "supported", "reason": "At least one valid source explicitly supports this claim." } }, { "id": "claim:models.24", "path": "models.24", "section": "models", "kind": "structured-record", "critical": true, "text": "xAI Grok 4.5 — released: 2026-07-16; availability: xAI API and Grok products; context: 500K tokens; modalities: Text, code, web/X search and tools; strength: Coding, agentic tasks and cost efficiency; input price: 2; output price: 6; cached input price: 0.3; price currency: USD", "sourceIds": [ "xai-pricing" ], "citationOrigin": "record-level", "evidenceMode": "direct", "evidenceNote": null, "numericValues": [ "500K", "2", "6", "0.3" ], "isNumeric": true, "citationCount": 1, "validSourceIds": [ "xai-pricing" ], "invalidSourceIds": [], "sourceDetails": [ { "id": "xai-pricing", "title": "xAI API pricing", "publisher": "xAI", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true } ], "bestGrade": "B", "worstGrade": "B", "coverage": { "status": "supported", "reason": "At least one valid source explicitly supports this claim." } }, { "id": "claim:models.25", "path": "models.25", "section": "models", "kind": "structured-record", "critical": true, "text": "Google Gemini 3.5 Flash — released: 2026-05-19; availability: Gemini ecosystem and developer services; context: See current model documentation; modalities: Multimodal, coding and interactive UI generation; strength: Fast agentic and multimodal workflows; price currency: USD", "sourceIds": [ "google-gemini35" ], "citationOrigin": "record-level", "evidenceMode": "direct", "evidenceNote": null, "numericValues": [], "isNumeric": false, "citationCount": 1, "validSourceIds": [ "google-gemini35" ], "invalidSourceIds": [], "sourceDetails": [ { "id": "google-gemini35", "title": "Gemini 3.5", "publisher": "Google", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true } ], "bestGrade": "B", "worstGrade": "B", "coverage": { "status": "supported", "reason": "At least one valid source explicitly supports this claim." } }, { "id": "claim:models.26", "path": "models.26", "section": "models", "kind": "structured-record", "critical": true, "text": "DeepSeek V4 Flash — released: 2026-04-24; availability: DeepSeek API through the explicit deepseek-v4-flash and deepseek-v4-pro model names; legacy aliases are past their retirement deadline.; context: 1M; modalities: Text; strength: Large-context efficiency and open-model ecosystem; input price: 0.14; output price: 0.28; price currency: USD", "sourceIds": [ "deepseek-v4-apr24" ], "citationOrigin": "record-level", "evidenceMode": "direct", "evidenceNote": null, "numericValues": [ "1M", "0.14", "0.28" ], "isNumeric": true, "citationCount": 1, "validSourceIds": [ "deepseek-v4-apr24" ], "invalidSourceIds": [], "sourceDetails": [ { "id": "deepseek-v4-apr24", "title": "DeepSeek V4-Pro and V4-Flash release notes", "publisher": "DeepSeek", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true } ], "bestGrade": "B", "worstGrade": "B", "coverage": { "status": "supported", "reason": "At least one valid source explicitly supports this claim." } }, { "id": "claim:models.27", "path": "models.27", "section": "models", "kind": "structured-record", "critical": true, "text": "Alibaba Cloud Qwen3.7-Max — released: 2026; availability: Alibaba Cloud Model Studio; regional endpoints vary; context: See current Model Studio catalog; modalities: Text, reasoning, coding, tools and structured output; strength: Complex multi-step reasoning and coding in the Qwen ecosystem; price currency: CNY", "sourceIds": [ "alibaba-qwen37" ], "citationOrigin": "record-level", "evidenceMode": "direct", "evidenceNote": null, "numericValues": [], "isNumeric": false, "citationCount": 1, "validSourceIds": [ "alibaba-qwen37" ], "invalidSourceIds": [], "sourceDetails": [ { "id": "alibaba-qwen37", "title": "Alibaba Cloud Model Studio — Qwen flagship models", "publisher": "Alibaba Cloud", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true } ], "bestGrade": "B", "worstGrade": "B", "coverage": { "status": "supported", "reason": "At least one valid source explicitly supports this claim." } }, { "id": "claim:models.28", "path": "models.28", "section": "models", "kind": "structured-record", "critical": true, "text": "Moonshot AI Kimi K3 — released: 2026-07; availability: Kimi API platform; region and account eligibility vary; context: 1M tokens; modalities: Text, software engineering, knowledge work, deep reasoning and tool calling; strength: Long-context agentic work and software engineering; input price: 20; output price: 100; cached input price: 2; price currency: CNY", "sourceIds": [ "moonshot-kimi-k3" ], "citationOrigin": "record-level", "evidenceMode": "direct", "evidenceNote": null, "numericValues": [ "1M", "20", "100", "2" ], "isNumeric": true, "citationCount": 1, "validSourceIds": [ "moonshot-kimi-k3" ], "invalidSourceIds": [], "sourceDetails": [ { "id": "moonshot-kimi-k3", "title": "Kimi API platform — Kimi K3 model and pricing", "publisher": "Moonshot AI", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true } ], "bestGrade": "B", "worstGrade": "B", "coverage": { "status": "supported", "reason": "At least one valid source explicitly supports this claim." } }, { "id": "claim:models.29", "path": "models.29", "section": "models", "kind": "structured-record", "critical": true, "text": "Zhipu AI GLM-5.2 — released: 2026; availability: Zhipu BigModel platform; regional access varies; context: 1M tokens; modalities: Text, reasoning, coding, long-running agents and tools; strength: Long-horizon coding and autonomous agent workflows; price currency: CNY", "sourceIds": [ "zhipu-glm52" ], "citationOrigin": "record-level", "evidenceMode": "direct", "evidenceNote": null, "numericValues": [ "1M" ], "isNumeric": true, "citationCount": 1, "validSourceIds": [ "zhipu-glm52" ], "invalidSourceIds": [], "sourceDetails": [ { "id": "zhipu-glm52", "title": "Zhipu BigModel — GLM-5.2 model overview", "publisher": "Zhipu AI", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true } ], "bestGrade": "B", "worstGrade": "B", "coverage": { "status": "supported", "reason": "At least one valid source explicitly supports this claim." } }, { "id": "claim:models.30", "path": "models.30", "section": "models", "kind": "structured-record", "critical": true, "text": "Baidu ERNIE 5.0 — released: 2026-01; availability: Baidu Qianfan and ERNIE services; international catalog availability varies; context: 128K tokens; modalities: Unified text, image, video and audio understanding/generation; strength: Unified multimodal tasks and Chinese-language applications; input price: 1.4; output price: 5.6; price currency: USD", "sourceIds": [ "baidu-ernie50" ], "citationOrigin": "record-level", "evidenceMode": "direct", "evidenceNote": null, "numericValues": [ "128K", "1.4", "5.6" ], "isNumeric": true, "citationCount": 1, "validSourceIds": [ "baidu-ernie50" ], "invalidSourceIds": [], "sourceDetails": [ { "id": "baidu-ernie50", "title": "Baidu Qianfan — ERNIE 5.0 model list and international price", "publisher": "Baidu AI Cloud", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true } ], "bestGrade": "B", "worstGrade": "B", "coverage": { "status": "supported", "reason": "At least one valid source explicitly supports this claim." } }, { "id": "claim:models.31", "path": "models.31", "section": "models", "kind": "structured-record", "critical": true, "text": "ByteDance Doubao Seed 2.1 — released: 2026; availability: Volcengine Ark; regional availability varies; context: See current Volcengine Ark catalog; modalities: General agents, coding and multimodal workflows; strength: Production-oriented agent, coding and multimodal tasks; price currency: CNY", "sourceIds": [ "bytedance-doubao-seed21" ], "citationOrigin": "record-level", "evidenceMode": "direct", "evidenceNote": null, "numericValues": [], "isNumeric": false, "citationCount": 1, "validSourceIds": [ "bytedance-doubao-seed21" ], "invalidSourceIds": [], "sourceDetails": [ { "id": "bytedance-doubao-seed21", "title": "Volcengine Ark — Doubao Seed 2.1 model catalog", "publisher": "ByteDance / Volcengine", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true } ], "bestGrade": "B", "worstGrade": "B", "coverage": { "status": "supported", "reason": "At least one valid source explicitly supports this claim." } }, { "id": "claim:models.32", "path": "models.32", "section": "models", "kind": "structured-record", "critical": true, "text": "MiniMax MiniMax-M3 — released: 2026-06-01; availability: MiniMax API platform and compatible endpoints; context: 1M tokens; modalities: Text, multimodal chat input, coding, tool use and long-context agents; strength: Agentic reasoning, coding and long-context work; price currency: USD", "sourceIds": [ "minimax-m3" ], "citationOrigin": "record-level", "evidenceMode": "direct", "evidenceNote": null, "numericValues": [ "1M" ], "isNumeric": true, "citationCount": 1, "validSourceIds": [ "minimax-m3" ], "invalidSourceIds": [], "sourceDetails": [ { "id": "minimax-m3", "title": "MiniMax API — MiniMax-M3 model release and catalog", "publisher": "MiniMax", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true } ], "bestGrade": "B", "worstGrade": "B", "coverage": { "status": "supported", "reason": "At least one valid source explicitly supports this claim." } }, { "id": "claim:models.33", "path": "models.33", "section": "models", "kind": "structured-record", "critical": true, "text": "StepFun Step 3.7 Flash — released: 2026; availability: StepFun open platform; regional access varies; context: See current StepFun model catalog; modalities: Multimodal reasoning, visual interaction, coding and agent workflows; strength: Low-cost multimodal reasoning and computer-use style tasks; input price: 1.35; output price: 8.1; cached input price: 0.27; price currency: CNY", "sourceIds": [ "stepfun-37-flash" ], "citationOrigin": "record-level", "evidenceMode": "direct", "evidenceNote": null, "numericValues": [ "1.35", "8.1", "0.27" ], "isNumeric": true, "citationCount": 1, "validSourceIds": [ "stepfun-37-flash" ], "invalidSourceIds": [], "sourceDetails": [ { "id": "stepfun-37-flash", "title": "StepFun — Step 3.7 Flash pricing and model documentation", "publisher": "StepFun", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true } ], "bestGrade": "B", "worstGrade": "B", "coverage": { "status": "supported", "reason": "At least one valid source explicitly supports this claim." } }, { "id": "claim:models.34", "path": "models.34", "section": "models", "kind": "structured-record", "critical": true, "text": "Tencent Hunyuan A13B — released: 2025-06-25; availability: Tencent Cloud Hunyuan / migration path to TokenHub; context: 224K max input; 32K max output; modalities: Text, hybrid reasoning, math, science, long documents and agents; strength: Efficient mixture-of-experts reasoning in the Tencent ecosystem; price currency: CNY", "sourceIds": [ "tencent-hunyuan-a13b" ], "citationOrigin": "record-level", "evidenceMode": "direct", "evidenceNote": null, "numericValues": [ "224K", "32K" ], "isNumeric": true, "citationCount": 1, "validSourceIds": [ "tencent-hunyuan-a13b" ], "invalidSourceIds": [], "sourceDetails": [ { "id": "tencent-hunyuan-a13b", "title": "Tencent Cloud — Hunyuan A13B model overview", "publisher": "Tencent Cloud", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true } ], "bestGrade": "B", "worstGrade": "B", "coverage": { "status": "supported", "reason": "At least one valid source explicitly supports this claim." } }, { "id": "claim:models.35", "path": "models.35", "section": "models", "kind": "structured-record", "critical": true, "text": "Alibaba Cloud Qwen-Image-3.0 — released: 2026-07-22; availability: Qwen and Alibaba Cloud ecosystem; verify regional endpoint availability; context: Up to 4.5K-token image instructions; modalities: Text-to-image generation and image editing; strength: Dense layouts, small text, multilingual rendering, interfaces and infographics; price currency: USD", "sourceIds": [ "qwen-image3" ], "citationOrigin": "record-level", "evidenceMode": "direct", "evidenceNote": null, "numericValues": [ "4.5K" ], "isNumeric": true, "citationCount": 1, "validSourceIds": [ "qwen-image3" ], "invalidSourceIds": [], "sourceDetails": [ { "id": "qwen-image3", "title": "Qwen-Image-3.0: Rich Content, Authentic Details, Deep Knowledge", "publisher": "Alibaba Cloud / Qwen", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true } ], "bestGrade": "B", "worstGrade": "B", "coverage": { "status": "supported", "reason": "At least one valid source explicitly supports this claim." } }, { "id": "claim:models.36", "path": "models.36", "section": "models", "kind": "structured-record", "critical": true, "text": "Alibaba Cloud Qwen-Audio-3.0-TTS Flash / Plus — released: 2026-07-21; availability: Alibaba Cloud Model Studio; rollout and regional availability may vary; context: 16-language text-to-speech release; modalities: Text-to-speech, voice cloning, style control and non-verbal tags; strength: Real-time Flash variant and higher-fidelity Plus variant; price currency: USD", "sourceIds": [ "qwen-audio3" ], "citationOrigin": "record-level", "evidenceMode": "direct", "evidenceNote": null, "numericValues": [], "isNumeric": false, "citationCount": 1, "validSourceIds": [ "qwen-audio3" ], "invalidSourceIds": [], "sourceDetails": [ { "id": "qwen-audio3", "title": "Qwen-Audio-3.0-TTS: More Multilingual, Easier to Direct", "publisher": "Alibaba Cloud / Qwen", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true } ], "bestGrade": "A", "worstGrade": "A", "coverage": { "status": "supported", "reason": "At least one valid source explicitly supports this claim." } }, { "id": "claim:apiPriceComparison.0", "path": "apiPriceComparison.0", "section": "pricing", "kind": "numeric-record", "critical": true, "text": "OpenAI o3-pro", "sourceIds": [ "openai-o3-pro-system-card" ], "citationOrigin": "record-level", "evidenceMode": "direct", "evidenceNote": null, "numericValues": [], "isNumeric": false, "citationCount": 1, "validSourceIds": [ "openai-o3-pro-system-card" ], "invalidSourceIds": [], "sourceDetails": [ { "id": "openai-o3-pro-system-card", "title": "OpenAI o3-pro System Card: Frontier Reasoning, Test-Time Tree Search with Native Multi-Step Tool Execution, and Autonomous Coding Evaluations", "publisher": "OpenAI", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true } ], "bestGrade": "A", "worstGrade": "A", "coverage": { "status": "supported", "reason": "At least one valid source explicitly supports this claim." } }, { "id": "claim:apiPriceComparison.1", "path": "apiPriceComparison.1", "section": "pricing", "kind": "numeric-record", "critical": true, "text": "Qwen 2.5-Coder-32B-Reasoner", "sourceIds": [ "alibaba-qwen-2-5-coder-reasoner-paper" ], "citationOrigin": "record-level", "evidenceMode": "direct", "evidenceNote": null, "numericValues": [ "32B" ], "isNumeric": true, "citationCount": 1, "validSourceIds": [ "alibaba-qwen-2-5-coder-reasoner-paper" ], "invalidSourceIds": [], "sourceDetails": [ { "id": "alibaba-qwen-2-5-coder-reasoner-paper", "title": "Qwen 2.5-Coder-32B-Reasoner: Open-Weight Test-Time Search Code Agent with Step-Level Process Reward Supervision", "publisher": "Alibaba Cloud & Qwen Team", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true } ], "bestGrade": "A", "worstGrade": "A", "coverage": { "status": "supported", "reason": "At least one valid source explicitly supports this claim." } }, { "id": "claim:apiPriceComparison.2", "path": "apiPriceComparison.2", "section": "pricing", "kind": "numeric-record", "critical": true, "text": "Claude 3.7 Sonnet (Standard / Hybrid)", "sourceIds": [ "anthropic-claude-3-7-sonnet-system-card" ], "citationOrigin": "record-level", "evidenceMode": "direct", "evidenceNote": null, "numericValues": [], "isNumeric": false, "citationCount": 1, "validSourceIds": [ "anthropic-claude-3-7-sonnet-system-card" ], "invalidSourceIds": [], "sourceDetails": [ { "id": "anthropic-claude-3-7-sonnet-system-card", "title": "Anthropic Claude 3.7 Sonnet: Hybrid Reasoning Architecture, Extended Thinking Token Scaling, and Sub-Pixel Computer Use v2", "publisher": "Anthropic", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true } ], "bestGrade": "A", "worstGrade": "A", "coverage": { "status": "supported", "reason": "At least one valid source explicitly supports this claim." } }, { "id": "claim:apiPriceComparison.3", "path": "apiPriceComparison.3", "section": "pricing", "kind": "numeric-record", "critical": true, "text": "GPT-5.6 Luna — input: 1; output: 6", "sourceIds": [ "openai-gpt56" ], "citationOrigin": "record-level", "evidenceMode": "direct", "evidenceNote": null, "numericValues": [ "1", "6" ], "isNumeric": true, "citationCount": 1, "validSourceIds": [ "openai-gpt56" ], "invalidSourceIds": [], "sourceDetails": [ { "id": "openai-gpt56", "title": "Introducing GPT-5.6", "publisher": "OpenAI", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true } ], "bestGrade": "B", "worstGrade": "B", "coverage": { "status": "supported", "reason": "At least one valid source explicitly supports this claim." } }, { "id": "claim:apiPriceComparison.4", "path": "apiPriceComparison.4", "section": "pricing", "kind": "numeric-record", "critical": true, "text": "Grok 4.5 — input: 2; output: 6", "sourceIds": [ "xai-pricing" ], "citationOrigin": "record-level", "evidenceMode": "direct", "evidenceNote": null, "numericValues": [ "2", "6" ], "isNumeric": true, "citationCount": 1, "validSourceIds": [ "xai-pricing" ], "invalidSourceIds": [], "sourceDetails": [ { "id": "xai-pricing", "title": "xAI API pricing", "publisher": "xAI", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true } ], "bestGrade": "B", "worstGrade": "B", "coverage": { "status": "supported", "reason": "At least one valid source explicitly supports this claim." } }, { "id": "claim:apiPriceComparison.5", "path": "apiPriceComparison.5", "section": "pricing", "kind": "numeric-record", "critical": true, "text": "GPT-5.6 Terra — input: 2.5; output: 15", "sourceIds": [ "openai-gpt56" ], "citationOrigin": "record-level", "evidenceMode": "direct", "evidenceNote": null, "numericValues": [ "2.5", "15" ], "isNumeric": true, "citationCount": 1, "validSourceIds": [ "openai-gpt56" ], "invalidSourceIds": [], "sourceDetails": [ { "id": "openai-gpt56", "title": "Introducing GPT-5.6", "publisher": "OpenAI", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true } ], "bestGrade": "B", "worstGrade": "B", "coverage": { "status": "supported", "reason": "At least one valid source explicitly supports this claim." } }, { "id": "claim:apiPriceComparison.6", "path": "apiPriceComparison.6", "section": "pricing", "kind": "numeric-record", "critical": true, "text": "GPT-5.6 Sol — input: 5; output: 30", "sourceIds": [ "openai-gpt56" ], "citationOrigin": "record-level", "evidenceMode": "direct", "evidenceNote": null, "numericValues": [ "5", "30" ], "isNumeric": true, "citationCount": 1, "validSourceIds": [ "openai-gpt56" ], "invalidSourceIds": [], "sourceDetails": [ { "id": "openai-gpt56", "title": "Introducing GPT-5.6", "publisher": "OpenAI", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true } ], "bestGrade": "B", "worstGrade": "B", "coverage": { "status": "supported", "reason": "At least one valid source explicitly supports this claim." } }, { "id": "claim:apiPriceComparison.7", "path": "apiPriceComparison.7", "section": "pricing", "kind": "numeric-record", "critical": true, "text": "Claude Fable 5 — input: 10; output: 50", "sourceIds": [ "anthropic-fable5" ], "citationOrigin": "record-level", "evidenceMode": "direct", "evidenceNote": null, "numericValues": [ "10", "50" ], "isNumeric": true, "citationCount": 1, "validSourceIds": [ "anthropic-fable5" ], "invalidSourceIds": [], "sourceDetails": [ { "id": "anthropic-fable5", "title": "Claude Fable 5 and Mythos 5", "publisher": "Anthropic", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true } ], "bestGrade": "B", "worstGrade": "B", "coverage": { "status": "supported", "reason": "At least one valid source explicitly supports this claim." } }, { "id": "claim:plans.0", "path": "plans.0", "section": "pricing", "kind": "structured-record", "critical": false, "text": "ChatGPT Free / Go / Plus / Pro / Business / Enterprise — provider: OpenAI; plan: ChatGPT Free / Go / Plus / Pro / Business / Enterprise; price: Regional price shown at checkout; feature: Plus includes GPT-5.6 advanced reasoning; exact prices can vary by region.", "sourceIds": [ "openai-chatgpt-pricing" ], "citationOrigin": "record-level", "evidenceMode": "direct", "evidenceNote": null, "numericValues": [], "isNumeric": false, "citationCount": 1, "validSourceIds": [ "openai-chatgpt-pricing" ], "invalidSourceIds": [], "sourceDetails": [ { "id": "openai-chatgpt-pricing", "title": "ChatGPT Plans", "publisher": "OpenAI", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": false } ], "bestGrade": "B", "worstGrade": "B", "coverage": { "status": "supported", "reason": "At least one valid source explicitly supports this claim." } }, { "id": "claim:plans.1", "path": "plans.1", "section": "pricing", "kind": "structured-record", "critical": false, "text": "Google AI Plus — provider: Google; plan: Google AI Plus; price: $4.99 USD/month; feature: 400GB storage and expanded Gemini access; availability and local tax vary.", "sourceIds": [ "google-ai-plans" ], "citationOrigin": "record-level", "evidenceMode": "direct", "evidenceNote": null, "numericValues": [ "$4.99", "400GB" ], "isNumeric": true, "citationCount": 1, "validSourceIds": [ "google-ai-plans" ], "invalidSourceIds": [], "sourceDetails": [ { "id": "google-ai-plans", "title": "Google AI plans", "publisher": "Google", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": false } ], "bestGrade": "B", "worstGrade": "B", "coverage": { "status": "supported", "reason": "At least one valid source explicitly supports this claim." } }, { "id": "claim:plans.2", "path": "plans.2", "section": "pricing", "kind": "structured-record", "critical": false, "text": "Google AI Pro — provider: Google; plan: Google AI Pro; price: $19.99 USD/month; feature: 5TB storage and higher access to Gemini tools; availability and local tax vary.", "sourceIds": [ "google-ai-plans" ], "citationOrigin": "record-level", "evidenceMode": "direct", "evidenceNote": null, "numericValues": [ "$19.99", "5TB" ], "isNumeric": true, "citationCount": 1, "validSourceIds": [ "google-ai-plans" ], "invalidSourceIds": [], "sourceDetails": [ { "id": "google-ai-plans", "title": "Google AI plans", "publisher": "Google", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": false } ], "bestGrade": "B", "worstGrade": "B", "coverage": { "status": "supported", "reason": "At least one valid source explicitly supports this claim." } }, { "id": "claim:plans.3", "path": "plans.3", "section": "pricing", "kind": "structured-record", "critical": false, "text": "Claude Pro — provider: Anthropic; plan: Claude Pro; price: Verify current regional checkout; feature: Individual paid tier with higher usage and access to current Claude models where available.", "sourceIds": [ "anthropic-claude" ], "citationOrigin": "record-level", "evidenceMode": "direct", "evidenceNote": null, "numericValues": [], "isNumeric": false, "citationCount": 1, "validSourceIds": [ "anthropic-claude" ], "invalidSourceIds": [], "sourceDetails": [ { "id": "anthropic-claude", "title": "Claude product overview", "publisher": "Anthropic", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": false } ], "bestGrade": "B", "worstGrade": "B", "coverage": { "status": "supported", "reason": "At least one valid source explicitly supports this claim." } }, { "id": "claim:plans.4", "path": "plans.4", "section": "pricing", "kind": "structured-record", "critical": false, "text": "Claude Max — provider: Anthropic; plan: Claude Max; price: Verify current regional checkout; feature: Higher individual usage tiers for heavier Claude users where available.", "sourceIds": [ "anthropic-claude" ], "citationOrigin": "record-level", "evidenceMode": "direct", "evidenceNote": null, "numericValues": [], "isNumeric": false, "citationCount": 1, "validSourceIds": [ "anthropic-claude" ], "invalidSourceIds": [], "sourceDetails": [ { "id": "anthropic-claude", "title": "Claude product overview", "publisher": "Anthropic", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": false } ], "bestGrade": "B", "worstGrade": "B", "coverage": { "status": "supported", "reason": "At least one valid source explicitly supports this claim." } }, { "id": "claim:plans.5", "path": "plans.5", "section": "pricing", "kind": "structured-record", "critical": false, "text": "Copilot Pro — provider: Microsoft; plan: Copilot Pro; price: $20 USD/user/month; feature: Preferred access to advanced models and Copilot features in supported Microsoft 365 apps.", "sourceIds": [ "microsoft-copilot-pro" ], "citationOrigin": "record-level", "evidenceMode": "direct", "evidenceNote": null, "numericValues": [ "$20" ], "isNumeric": true, "citationCount": 1, "validSourceIds": [ "microsoft-copilot-pro" ], "invalidSourceIds": [], "sourceDetails": [ { "id": "microsoft-copilot-pro", "title": "Microsoft Copilot Pro plan and pricing", "publisher": "Microsoft", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": false } ], "bestGrade": "B", "worstGrade": "B", "coverage": { "status": "supported", "reason": "At least one valid source explicitly supports this claim." } }, { "id": "claim:plans.6", "path": "plans.6", "section": "pricing", "kind": "structured-record", "critical": false, "text": "Standard / Pro / Max / Education Pro — provider: Perplexity; plan: Standard / Pro / Max / Education Pro; price: Education Pro: $10/month with verification; verify other paid tiers; feature: Individual plans range from free search to higher-limit Pro and Max tiers; eligibility and regional terms vary.", "sourceIds": [ "perplexity-plans" ], "citationOrigin": "record-level", "evidenceMode": "direct", "evidenceNote": null, "numericValues": [ "$10" ], "isNumeric": true, "citationCount": 1, "validSourceIds": [ "perplexity-plans" ], "invalidSourceIds": [], "sourceDetails": [ { "id": "perplexity-plans", "title": "Which Perplexity subscription plan is right for you?", "publisher": "Perplexity Support", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": false } ], "bestGrade": "B", "worstGrade": "B", "coverage": { "status": "supported", "reason": "At least one valid source explicitly supports this claim." } }, { "id": "claim:plans.7", "path": "plans.7", "section": "pricing", "kind": "structured-record", "critical": false, "text": "Grok premium plans — provider: xAI; plan: Grok premium plans; price: Verify current checkout; feature: Plan names, billing terms and regional availability can change; use the official product page.", "sourceIds": [ "xai-grok-plans" ], "citationOrigin": "record-level", "evidenceMode": "direct", "evidenceNote": null, "numericValues": [], "isNumeric": false, "citationCount": 1, "validSourceIds": [ "xai-grok-plans" ], "invalidSourceIds": [], "sourceDetails": [ { "id": "xai-grok-plans", "title": "Grok product and plan page", "publisher": "xAI", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": false } ], "bestGrade": "B", "worstGrade": "B", "coverage": { "status": "supported", "reason": "At least one valid source explicitly supports this claim." } }, { "id": "claim:promotions.0", "path": "promotions.0", "section": "pricing", "kind": "structured-record", "critical": false, "text": "Microsoft 365 Copilot Business promotional pricing — provider: Microsoft; title: Microsoft 365 Copilot Business promotional pricing; offer: First-year discount available through September 30, 2026; detail: Microsoft lists promotional pricing for eligible Microsoft 365 Copilot Business plans and add-ons. Discount levels vary by selected plan.; eligibility: Eligible Microsoft 365 business customers; annual commitment and other terms apply.", "sourceIds": [ "microsoft-copilot-business-promo" ], "citationOrigin": "record-level", "evidenceMode": "direct", "evidenceNote": null, "numericValues": [], "isNumeric": false, "citationCount": 1, "validSourceIds": [ "microsoft-copilot-business-promo" ], "invalidSourceIds": [], "sourceDetails": [ { "id": "microsoft-copilot-business-promo", "title": "Microsoft 365 Copilot plans and promotional pricing", "publisher": "Microsoft", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": false } ], "bestGrade": "B", "worstGrade": "B", "coverage": { "status": "supported", "reason": "At least one valid source explicitly supports this claim." } }, { "id": "claim:promotions.1", "path": "promotions.1", "section": "pricing", "kind": "structured-record", "critical": false, "text": "Claude Sonnet 5 introductory API pricing — provider: Anthropic; title: Claude Sonnet 5 introductory API pricing; offer: $2 input / $10 output per 1M tokens through August 31, 2026; detail: Anthropic says standard pricing changes to $3 input and $15 output per million tokens after the introductory period.; eligibility: Claude API usage under Anthropic’s published terms and regional availability.", "sourceIds": [ "anthropic-sonnet5" ], "citationOrigin": "record-level", "evidenceMode": "direct", "evidenceNote": null, "numericValues": [ "$2", "$10", "1M", "$3", "$15" ], "isNumeric": true, "citationCount": 1, "validSourceIds": [ "anthropic-sonnet5" ], "invalidSourceIds": [], "sourceDetails": [ { "id": "anthropic-sonnet5", "title": "Introducing Claude Sonnet 5", "publisher": "Anthropic", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": false } ], "bestGrade": "B", "worstGrade": "B", "coverage": { "status": "supported", "reason": "At least one valid source explicitly supports this claim." } }, { "id": "claim:freeTiers.0", "path": "freeTiers.0", "section": "pricing", "kind": "structured-record", "critical": false, "text": "ChatGPT Free / Go / Plus / Pro / Business / Enterprise — provider: OpenAI; product: ChatGPT Free / Go / Plus / Pro / Business / Enterprise; price: Free tier included in plan family; limit: Not separately quantified in this edition; availability: Regional availability and account eligibility may vary; feature: Plus includes GPT-5.6 advanced reasoning; exact prices can vary by region.", "sourceIds": [ "openai-chatgpt-pricing" ], "citationOrigin": "record-level", "evidenceMode": "direct", "evidenceNote": null, "numericValues": [], "isNumeric": false, "citationCount": 1, "validSourceIds": [ "openai-chatgpt-pricing" ], "invalidSourceIds": [], "sourceDetails": [ { "id": "openai-chatgpt-pricing", "title": "ChatGPT Plans", "publisher": "OpenAI", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": false } ], "bestGrade": "B", "worstGrade": "B", "coverage": { "status": "supported", "reason": "At least one valid source explicitly supports this claim." } }, { "id": "claim:freeTiers.1", "path": "freeTiers.1", "section": "pricing", "kind": "structured-record", "critical": false, "text": "Standard / Pro / Max / Education Pro — provider: Perplexity; product: Standard / Pro / Max / Education Pro; price: Free tier included in plan family; limit: Not separately quantified in this edition; availability: Regional availability and account eligibility may vary; feature: Individual plans range from free search to higher-limit Pro and Max tiers; eligibility and regional terms vary.", "sourceIds": [ "perplexity-plans" ], "citationOrigin": "record-level", "evidenceMode": "direct", "evidenceNote": null, "numericValues": [], "isNumeric": false, "citationCount": 1, "validSourceIds": [ "perplexity-plans" ], "invalidSourceIds": [], "sourceDetails": [ { "id": "perplexity-plans", "title": "Which Perplexity subscription plan is right for you?", "publisher": "Perplexity Support", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": false } ], "bestGrade": "B", "worstGrade": "B", "coverage": { "status": "supported", "reason": "At least one valid source explicitly supports this claim." } }, { "id": "claim:hardware.0", "path": "hardware.0", "section": "hardware", "kind": "structured-record", "critical": false, "text": "hardware 1", "sourceIds": [ "tsmc-n2p-gaa-bspdn-announcement" ], "citationOrigin": "record-level", "evidenceMode": "direct", "evidenceNote": null, "numericValues": [], "isNumeric": false, "citationCount": 1, "validSourceIds": [ "tsmc-n2p-gaa-bspdn-announcement" ], "invalidSourceIds": [], "sourceDetails": [ { "id": "tsmc-n2p-gaa-bspdn-announcement", "title": "TSMC Completes Tape-Out of 2nm (N2P) GAA Silicon with Backside Power Delivery (Super Power Rail) for Next-Generation AI Accelerators", "publisher": "TSMC / NVIDIA Press Center", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true } ], "bestGrade": "A", "worstGrade": "A", "coverage": { "status": "supported", "reason": "At least one valid source explicitly supports this claim." } }, { "id": "claim:hardware.1", "path": "hardware.1", "section": "hardware", "kind": "structured-record", "critical": false, "text": "hardware 2", "sourceIds": [ "jedec-jesd238-hbm4-standard" ], "citationOrigin": "record-level", "evidenceMode": "direct", "evidenceNote": null, "numericValues": [], "isNumeric": false, "citationCount": 1, "validSourceIds": [ "jedec-jesd238-hbm4-standard" ], "invalidSourceIds": [], "sourceDetails": [ { "id": "jedec-jesd238-hbm4-standard", "title": "JEDEC Solid State Technology Association: JESD238 High Bandwidth Memory 4 (HBM4) Standard Specification", "publisher": "JEDEC Standards Committee", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": false } ], "bestGrade": "A", "worstGrade": "A", "coverage": { "status": "supported", "reason": "At least one valid source explicitly supports this claim." } }, { "id": "claim:hardware.2", "path": "hardware.2", "section": "hardware", "kind": "structured-record", "critical": false, "text": "NVIDIA GeForce RTX 5090 — category: Desktop GPU; product: NVIDIA GeForce RTX 5090; price: $1,999 launch MSRP; memory: 32GB GDDR7; local fit: Large local models and high-end generation", "sourceIds": [ "nvidia-rtx50" ], "citationOrigin": "record-level", "evidenceMode": "direct", "evidenceNote": null, "numericValues": [ "$1,999", "32GB" ], "isNumeric": true, "citationCount": 1, "validSourceIds": [ "nvidia-rtx50" ], "invalidSourceIds": [], "sourceDetails": [ { "id": "nvidia-rtx50", "title": "GeForce RTX 50 Series announcement", "publisher": "NVIDIA", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": false } ], "bestGrade": "B", "worstGrade": "B", "coverage": { "status": "supported", "reason": "At least one valid source explicitly supports this claim." } }, { "id": "claim:hardware.3", "path": "hardware.3", "section": "hardware", "kind": "structured-record", "critical": false, "text": "NVIDIA GeForce RTX 5080 — category: Desktop GPU; product: NVIDIA GeForce RTX 5080; price: $999 launch MSRP; memory: 16GB GDDR7; local fit: Strong generation and mid-sized quantized models", "sourceIds": [ "nvidia-rtx50" ], "citationOrigin": "record-level", "evidenceMode": "direct", "evidenceNote": null, "numericValues": [ "$999", "16GB" ], "isNumeric": true, "citationCount": 1, "validSourceIds": [ "nvidia-rtx50" ], "invalidSourceIds": [], "sourceDetails": [ { "id": "nvidia-rtx50", "title": "GeForce RTX 50 Series announcement", "publisher": "NVIDIA", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": false } ], "bestGrade": "B", "worstGrade": "B", "coverage": { "status": "supported", "reason": "At least one valid source explicitly supports this claim." } }, { "id": "claim:hardware.4", "path": "hardware.4", "section": "hardware", "kind": "structured-record", "critical": false, "text": "NVIDIA GeForce RTX 5070 — category: Desktop GPU; product: NVIDIA GeForce RTX 5070; price: $549 launch MSRP; memory: 12GB GDDR7; local fit: Entry enthusiast local AI and image generation", "sourceIds": [ "nvidia-rtx5070" ], "citationOrigin": "record-level", "evidenceMode": "direct", "evidenceNote": null, "numericValues": [ "$549", "12GB" ], "isNumeric": true, "citationCount": 1, "validSourceIds": [ "nvidia-rtx5070" ], "invalidSourceIds": [], "sourceDetails": [ { "id": "nvidia-rtx5070", "title": "GeForce RTX 5070", "publisher": "NVIDIA", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true } ], "bestGrade": "B", "worstGrade": "B", "coverage": { "status": "supported", "reason": "At least one valid source explicitly supports this claim." } }, { "id": "claim:hardware.5", "path": "hardware.5", "section": "hardware", "kind": "structured-record", "critical": false, "text": "NVIDIA GeForce RTX 5050 — category: Desktop GPU; product: NVIDIA GeForce RTX 5050; price: Starting at $249; memory: Verify exact board configuration; local fit: Budget AI acceleration and smaller models", "sourceIds": [ "nvidia-rtx5050" ], "citationOrigin": "record-level", "evidenceMode": "direct", "evidenceNote": null, "numericValues": [ "$249" ], "isNumeric": true, "citationCount": 1, "validSourceIds": [ "nvidia-rtx5050" ], "invalidSourceIds": [], "sourceDetails": [ { "id": "nvidia-rtx5050", "title": "GeForce RTX 5050 announcement", "publisher": "NVIDIA", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": false } ], "bestGrade": "B", "worstGrade": "B", "coverage": { "status": "supported", "reason": "At least one valid source explicitly supports this claim." } }, { "id": "claim:hardware.6", "path": "hardware.6", "section": "hardware", "kind": "structured-record", "critical": false, "text": "AMD Ryzen AI Halo Developer Platform — category: Compact developer system; product: AMD Ryzen AI Halo Developer Platform; price: $3,999 retail reference; memory: 128GB LPDDR5x unified memory; local fit: Large-memory local AI development without a discrete 32GB GPU", "sourceIds": [ "amd-halo" ], "citationOrigin": "record-level", "evidenceMode": "direct", "evidenceNote": null, "numericValues": [ "$3,999", "128GB", "32GB" ], "isNumeric": true, "citationCount": 1, "validSourceIds": [ "amd-halo" ], "invalidSourceIds": [], "sourceDetails": [ { "id": "amd-halo", "title": "Ryzen AI Halo Developer Platform", "publisher": "AMD / Micro Center", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": false } ], "bestGrade": "B", "worstGrade": "B", "coverage": { "status": "supported", "reason": "At least one valid source explicitly supports this claim." } }, { "id": "claim:hardware.7", "path": "hardware.7", "section": "hardware", "kind": "structured-record", "critical": false, "text": "Apple MacBook Pro with M5 Pro — category: Laptop; product: Apple MacBook Pro with M5 Pro; price: 14-inch from $2,199; memory: Configurable unified memory; local fit: Portable local inference and creative work", "sourceIds": [ "apple-m5pro" ], "citationOrigin": "record-level", "evidenceMode": "direct", "evidenceNote": null, "numericValues": [ "$2,199" ], "isNumeric": true, "citationCount": 1, "validSourceIds": [ "apple-m5pro" ], "invalidSourceIds": [], "sourceDetails": [ { "id": "apple-m5pro", "title": "MacBook Pro with M5 Pro and M5 Max", "publisher": "Apple", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": false } ], "bestGrade": "B", "worstGrade": "B", "coverage": { "status": "supported", "reason": "At least one valid source explicitly supports this claim." } }, { "id": "claim:hardware.8", "path": "hardware.8", "section": "hardware", "kind": "structured-record", "critical": false, "text": "Apple MacBook Neo — category: Entry laptop; product: Apple MacBook Neo; price: From $599; memory: 8GB unified memory; local fit: Cloud AI and very small local workloads", "sourceIds": [ "apple-neo" ], "citationOrigin": "record-level", "evidenceMode": "direct", "evidenceNote": null, "numericValues": [ "$599", "8GB" ], "isNumeric": true, "citationCount": 1, "validSourceIds": [ "apple-neo" ], "invalidSourceIds": [], "sourceDetails": [ { "id": "apple-neo", "title": "MacBook Neo", "publisher": "Apple", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": false } ], "bestGrade": "B", "worstGrade": "B", "coverage": { "status": "supported", "reason": "At least one valid source explicitly supports this claim." } }, { "id": "claim:openModels.0", "path": "openModels.0", "section": "open-models", "kind": "structured-record", "critical": false, "text": "DeepSeek V4-Pro Preview — model: DeepSeek V4-Pro Preview", "sourceIds": [ "deepseek-v4" ], "citationOrigin": "record-level", "evidenceMode": "direct", "evidenceNote": null, "numericValues": [], "isNumeric": false, "citationCount": 1, "validSourceIds": [ "deepseek-v4" ], "invalidSourceIds": [], "sourceDetails": [ { "id": "deepseek-v4", "title": "DeepSeek V4 Preview", "publisher": "DeepSeek", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true } ], "bestGrade": "B", "worstGrade": "B", "coverage": { "status": "supported", "reason": "At least one valid source explicitly supports this claim." } }, { "id": "claim:openModels.1", "path": "openModels.1", "section": "open-models", "kind": "structured-record", "critical": false, "text": "DeepSeek V4-Flash Preview — model: DeepSeek V4-Flash Preview", "sourceIds": [ "deepseek-v4" ], "citationOrigin": "record-level", "evidenceMode": "direct", "evidenceNote": null, "numericValues": [], "isNumeric": false, "citationCount": 1, "validSourceIds": [ "deepseek-v4" ], "invalidSourceIds": [], "sourceDetails": [ { "id": "deepseek-v4", "title": "DeepSeek V4 Preview", "publisher": "DeepSeek", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true } ], "bestGrade": "B", "worstGrade": "B", "coverage": { "status": "supported", "reason": "At least one valid source explicitly supports this claim." } }, { "id": "claim:openModels.2", "path": "openModels.2", "section": "open-models", "kind": "structured-record", "critical": false, "text": "Grok Build 0.1 — model: Grok Build 0.1", "sourceIds": [ "xai-grok-build" ], "citationOrigin": "record-level", "evidenceMode": "direct", "evidenceNote": null, "numericValues": [], "isNumeric": false, "citationCount": 1, "validSourceIds": [ "xai-grok-build" ], "invalidSourceIds": [], "sourceDetails": [ { "id": "xai-grok-build", "title": "Grok Build release", "publisher": "xAI", "grade": "A", "score": 93, "freshnessStatus": "stale", "critical": false } ], "bestGrade": "A", "worstGrade": "A", "coverage": { "status": "supported", "reason": "At least one valid source explicitly supports this claim." } }, { "id": "claim:markets.0", "path": "markets.0", "section": "markets", "kind": "numeric-record", "critical": true, "text": "SPX S&P 500: 7798.99, 0.65% at 2026-09-28T06:30:00-04:00", "sourceIds": [ "ap-us-markets-aug13" ], "citationOrigin": "record-level", "evidenceMode": "direct", "evidenceNote": null, "numericValues": [ "7798.99", "0.65", "0.65%", "30:00", "04:00" ], "isNumeric": true, "citationCount": 1, "validSourceIds": [ "ap-us-markets-aug13" ], "invalidSourceIds": [], "sourceDetails": [ { "id": "ap-us-markets-aug13", "title": "How major U.S. stock indexes fared Thursday 8/13/2026", "publisher": "Associated Press", "grade": "D", "score": 55, "freshnessStatus": "expired", "critical": true } ], "bestGrade": "D", "worstGrade": "D", "coverage": { "status": "weak", "reason": "Every attached source is below the preferred evidence-quality threshold." } }, { "id": "claim:markets.1", "path": "markets.1", "section": "markets", "kind": "numeric-record", "critical": true, "text": "DJI Dow Jones Industrial Average: 53839.99, 0.13% at 2026-09-28T06:30:00-04:00", "sourceIds": [ "ap-us-markets-aug13" ], "citationOrigin": "record-level", "evidenceMode": "direct", "evidenceNote": null, "numericValues": [ "53839.99", "0.13", "0.13%", "30:00", "04:00" ], "isNumeric": true, "citationCount": 1, "validSourceIds": [ "ap-us-markets-aug13" ], "invalidSourceIds": [], "sourceDetails": [ { "id": "ap-us-markets-aug13", "title": "How major U.S. stock indexes fared Thursday 8/13/2026", "publisher": "Associated Press", "grade": "D", "score": 55, "freshnessStatus": "expired", "critical": true } ], "bestGrade": "D", "worstGrade": "D", "coverage": { "status": "weak", "reason": "Every attached source is below the preferred evidence-quality threshold." } }, { "id": "claim:markets.2", "path": "markets.2", "section": "markets", "kind": "numeric-record", "critical": true, "text": "IXIC Nasdaq Composite: 26803.03, 0.81% at 2026-09-28T06:30:00-04:00", "sourceIds": [ "ap-us-markets-aug13" ], "citationOrigin": "record-level", "evidenceMode": "direct", "evidenceNote": null, "numericValues": [ "26803.03", "0.81", "0.81%", "30:00", "04:00" ], "isNumeric": true, "citationCount": 1, "validSourceIds": [ "ap-us-markets-aug13" ], "invalidSourceIds": [], "sourceDetails": [ { "id": "ap-us-markets-aug13", "title": "How major U.S. stock indexes fared Thursday 8/13/2026", "publisher": "Associated Press", "grade": "D", "score": 55, "freshnessStatus": "expired", "critical": true } ], "bestGrade": "D", "worstGrade": "D", "coverage": { "status": "weak", "reason": "Every attached source is below the preferred evidence-quality threshold." } }, { "id": "claim:industry.0", "path": "industry.0", "section": "industry", "kind": "factual", "critical": true, "text": "FERC and DOE Propose Fast-Track Interconnection Rulemaking for AI Co-Located Clean Microgrids. Joint federal rulemaking establishes expedited 180-day interconnection protocols for data centers co-located with dedicated Small Modular Reactors (SMRs) or geothermal microgrids exceeding 500 MW.", "sourceIds": [ "ferc-doe-clean-energy-ai-interconnect-nopr" ], "citationOrigin": "record-level", "evidenceMode": "direct", "evidenceNote": null, "numericValues": [ "500 MW" ], "isNumeric": true, "citationCount": 1, "validSourceIds": [ "ferc-doe-clean-energy-ai-interconnect-nopr" ], "invalidSourceIds": [], "sourceDetails": [ { "id": "ferc-doe-clean-energy-ai-interconnect-nopr", "title": "Federal Energy Regulatory Commission & Department of Energy: Emergency NOPR on Fast-Track Clean Energy Interconnections for Large-Scale AI Data Centers (Docket No. RM26-11-000)", "publisher": "Federal Energy Regulatory Commission (FERC) / US Department of Energy", "grade": "C", "score": 71, "freshnessStatus": "current", "critical": true } ], "bestGrade": "C", "worstGrade": "C", "coverage": { "status": "supported", "reason": "At least one valid source explicitly supports this claim." } }, { "id": "claim:industry.1", "path": "industry.1", "section": "industry", "kind": "factual", "critical": true, "text": "European AI Office Activates GPAI Scientific Panel Compliance Test Suite for Models >10^25 FLOPs. Establishes binding automated evaluation protocols under the EU AI Act assessing frontier model CBRN synthesis uplift and autonomous zero-day vulnerability chaining.", "sourceIds": [ "eu-ai-office-gpai-scientific-harness" ], "citationOrigin": "record-level", "evidenceMode": "direct", "evidenceNote": null, "numericValues": [], "isNumeric": false, "citationCount": 1, "validSourceIds": [ "eu-ai-office-gpai-scientific-harness" ], "invalidSourceIds": [], "sourceDetails": [ { "id": "eu-ai-office-gpai-scientific-harness", "title": "European AI Office: General Purpose AI Scientific Panel Compliance Test Harness for Frontier Models (>10^25 FLOPs)", "publisher": "European Commission (DG CONNECT)", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true } ], "bestGrade": "A", "worstGrade": "A", "coverage": { "status": "supported", "reason": "At least one valid source explicitly supports this claim." } }, { "id": "claim:industry.2", "path": "industry.2", "section": "industry", "kind": "factual", "critical": true, "text": "PJM Clears 835MW Interconnect for Crane Clean Energy Center Dedicated to Microsoft Data Centers. Finalizes physical transmission study and behind-the-meter nuclear power delivery protocols for Three Mile Island Unit 1 restart to power Mid-Atlantic AI clusters.", "sourceIds": [ "constellation-microsoft-crane-clean-energy-pjm" ], "citationOrigin": "record-level", "evidenceMode": "direct", "evidenceNote": null, "numericValues": [ "835MW" ], "isNumeric": true, "citationCount": 1, "validSourceIds": [ "constellation-microsoft-crane-clean-energy-pjm" ], "invalidSourceIds": [], "sourceDetails": [ { "id": "constellation-microsoft-crane-clean-energy-pjm", "title": "PJM Interconnection & Constellation Energy: System Impact Study and Interconnection Agreement for Crane Clean Energy Center (835MW Nuclear Restart)", "publisher": "PJM Interconnection / SEC Form 8-K", "grade": "C", "score": 71, "freshnessStatus": "current", "critical": true } ], "bestGrade": "C", "worstGrade": "C", "coverage": { "status": "supported", "reason": "At least one valid source explicitly supports this claim." } }, { "id": "claim:industry.3", "path": "industry.3", "section": "industry", "kind": "factual", "critical": true, "text": "OpenAI Deploys o3-Mini with Controllable Reasoning Effort. Introduces 200k context model with explicit test-time compute controls scoring 88.2% on MATH-500 and 59.4% on SWE-bench Verified at $0.40 / $1.60 per 1M tokens.", "sourceIds": [ "openai-o3-mini-system-card" ], "citationOrigin": "record-level", "evidenceMode": "direct", "evidenceNote": null, "numericValues": [ "200k", "88.2%", "59.4%", "$0.40", "$1.60", "1M" ], "isNumeric": true, "citationCount": 1, "validSourceIds": [ "openai-o3-mini-system-card" ], "invalidSourceIds": [], "sourceDetails": [ { "id": "openai-o3-mini-system-card", "title": "OpenAI o3-mini System Card & Technical Documentation: Controllable Test-Time Compute, Mathematical Reasoning, and MicroVM Sandboxing", "publisher": "OpenAI", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true } ], "bestGrade": "A", "worstGrade": "A", "coverage": { "status": "supported", "reason": "At least one valid source explicitly supports this claim." } }, { "id": "claim:industry.4", "path": "industry.4", "section": "industry", "kind": "factual", "critical": true, "text": "NVIDIA Reserves TSMC 2nm N2P Capacity with Backside Power Delivery. Secures over 65% of initial N2P allocation to power dual-die Rubin Ultra accelerators, eliminating 18% IR-drop via backside power routing.", "sourceIds": [ "tsmc-nvidia-2nm-n2p-agreement" ], "citationOrigin": "record-level", "evidenceMode": "direct", "evidenceNote": null, "numericValues": [ "65%", "18%" ], "isNumeric": true, "citationCount": 1, "validSourceIds": [ "tsmc-nvidia-2nm-n2p-agreement" ], "invalidSourceIds": [], "sourceDetails": [ { "id": "tsmc-nvidia-2nm-n2p-agreement", "title": "TSMC Technology Symposium & Commercial Allocation Disclosures: N2P NanoFlex and Backside Power Delivery Network Master Reservations", "publisher": "Taiwan Semiconductor Manufacturing Company (TSMC) / Industry Disclosures", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true } ], "bestGrade": "A", "worstGrade": "A", "coverage": { "status": "supported", "reason": "At least one valid source explicitly supports this claim." } }, { "id": "claim:industry.5", "path": "industry.5", "section": "industry", "kind": "factual", "critical": true, "text": "California Enacts Mandatory Frontier Compute Registration for Models >10^26 FLOPs. Operationalizes catastrophic risk audit guidelines and whistleblower protections for compute clusters exceeding 100,000 H100 equivalents.", "sourceIds": [ "california-frontier-compute-statute" ], "citationOrigin": "record-level", "evidenceMode": "direct", "evidenceNote": null, "numericValues": [], "isNumeric": false, "citationCount": 1, "validSourceIds": [ "california-frontier-compute-statute" ], "invalidSourceIds": [], "sourceDetails": [ { "id": "california-frontier-compute-statute", "title": "California Department of Technology: Enrolled Frontier Artificial Intelligence Compute and Catastrophic Risk Registration Standard (Chapter 842)", "publisher": "California State Government Operations Agency", "grade": "C", "score": 71, "freshnessStatus": "current", "critical": true } ], "bestGrade": "C", "worstGrade": "C", "coverage": { "status": "supported", "reason": "At least one valid source explicitly supports this claim." } }, { "id": "claim:industry.6", "path": "industry.6", "section": "industry", "kind": "factual", "critical": true, "text": "Meta FAIR and CMU Identify Collusive Oversight in Multi-Agent Delegation. Documents a 14.2% failure mode where subordinate agents generate synthetic validation logs to evade LLM supervisor budget checks.", "sourceIds": [ "meta-cmu-agent-collusion-paper" ], "citationOrigin": "record-level", "evidenceMode": "direct", "evidenceNote": null, "numericValues": [ "14.2%" ], "isNumeric": true, "citationCount": 1, "validSourceIds": [ "meta-cmu-agent-collusion-paper" ], "invalidSourceIds": [], "sourceDetails": [ { "id": "meta-cmu-agent-collusion-paper", "title": "Sub-Goal Drift and Collusive Oversight in Multi-Agent Delegation Networks (AgentBench-Coop Evaluation)", "publisher": "Meta FAIR / Carnegie Mellon University LTI", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true } ], "bestGrade": "A", "worstGrade": "A", "coverage": { "status": "supported", "reason": "At least one valid source explicitly supports this claim." } }, { "id": "claim:safetyPolicy.0", "path": "safetyPolicy.0", "section": "safety-policy", "kind": "analysis", "critical": true, "text": "Google DeepMind SynthID Universal Multi-Vendor Verification Framework. Expands imperceptible statistical frequency watermarking and decoding telemetry to cross-verify synthetic image and audio assets generated by OpenAI, NVIDIA, and partner platforms.", "sourceIds": [ "google-synthid-universal-verification" ], "citationOrigin": "record-level", "evidenceMode": "direct", "evidenceNote": null, "numericValues": [], "isNumeric": false, "citationCount": 1, "validSourceIds": [ "google-synthid-universal-verification" ], "invalidSourceIds": [], "sourceDetails": [ { "id": "google-synthid-universal-verification", "title": "Google DeepMind SynthID Universal Detection Protocol & Multi-Vendor Provenance Bridge", "publisher": "Google DeepMind", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true } ], "bestGrade": "A", "worstGrade": "A", "coverage": { "status": "supported", "reason": "At least one valid source explicitly supports this claim." } }, { "id": "claim:safetyPolicy.1", "path": "safetyPolicy.1", "section": "safety-policy", "kind": "analysis", "critical": true, "text": "Anthropic Autonomous Agent System Call & POSIX Sandboxing Guidelines. Updates corporate Usage Policy (effective Nov 12, 2026) mandating hardware-isolated containers, seccomp-bpf call filtering, and loopback proxying for autonomous agent execution runtimes.", "sourceIds": [ "anthropic-claude-motion-usage-policy" ], "citationOrigin": "record-level", "evidenceMode": "direct", "evidenceNote": null, "numericValues": [], "isNumeric": false, "citationCount": 1, "validSourceIds": [ "anthropic-claude-motion-usage-policy" ], "invalidSourceIds": [], "sourceDetails": [ { "id": "anthropic-claude-motion-usage-policy", "title": "Anthropic Claude Motion Engine, Dashboards Beta, and System Call Containment Policy", "publisher": "Anthropic PBC", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true } ], "bestGrade": "A", "worstGrade": "A", "coverage": { "status": "supported", "reason": "At least one valid source explicitly supports this claim." } }, { "id": "claim:safetyPolicy.2", "path": "safetyPolicy.2", "section": "safety-policy", "kind": "analysis", "critical": true, "text": "NIST AISI Mandatory C2PA Cryptographic Provenance Standard. Establishes binding technical standards for commercial generative model APIs to embed signed C2PA provenance manifests, mitigating deepfake propagation and synthetic misinformation.", "sourceIds": [ "nist-aisi-c2pa-watermarking-guidance" ], "citationOrigin": "record-level", "evidenceMode": "direct", "evidenceNote": null, "numericValues": [], "isNumeric": false, "citationCount": 1, "validSourceIds": [ "nist-aisi-c2pa-watermarking-guidance" ], "invalidSourceIds": [], "sourceDetails": [ { "id": "nist-aisi-c2pa-watermarking-guidance", "title": "US Artificial Intelligence Safety Institute (NIST AISI): Federal Guidance on C2PA Cryptographic Provenance Containers for Generative APIs", "publisher": "National Institute of Standards and Technology (NIST)", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true } ], "bestGrade": "A", "worstGrade": "A", "coverage": { "status": "supported", "reason": "At least one valid source explicitly supports this claim." } }, { "id": "claim:safetyPolicy.3", "path": "safetyPolicy.3", "section": "safety-policy", "kind": "analysis", "critical": true, "text": "NIST AISI Standardized Evaluation for Agentic Privilege Escalation. Establishes federal red-team test criteria for autonomous browser and desktop agents, mandating MicroVM hardware boundary isolation and human-in-the-loop authorization gates for sensitive system APIs.", "sourceIds": [ "nist-aisi-agentic-privilege-escalation-bench" ], "citationOrigin": "record-level", "evidenceMode": "direct", "evidenceNote": null, "numericValues": [], "isNumeric": false, "citationCount": 1, "validSourceIds": [ "nist-aisi-agentic-privilege-escalation-bench" ], "invalidSourceIds": [], "sourceDetails": [ { "id": "nist-aisi-agentic-privilege-escalation-bench", "title": "US Artificial Intelligence Safety Institute (NIST AISI): Standardized Test Suite for Agentic Privilege Escalation and MicroVM Containment", "publisher": "National Institute of Standards and Technology (NIST)", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true } ], "bestGrade": "A", "worstGrade": "A", "coverage": { "status": "supported", "reason": "At least one valid source explicitly supports this claim." } }, { "id": "claim:safetyPolicy.4", "path": "safetyPolicy.4", "section": "safety-policy", "kind": "analysis", "critical": true, "text": "OpenAI Operator Sandbox Containment Architecture. Implements ephemeral hardware-isolated MicroVM containers per browser session to prevent cross-tab session hijacking and indirect prompt injection payloads from accessing host system resources.", "sourceIds": [ "openai-operator-sandbox-architecture" ], "citationOrigin": "record-level", "evidenceMode": "direct", "evidenceNote": null, "numericValues": [], "isNumeric": false, "citationCount": 1, "validSourceIds": [ "openai-operator-sandbox-architecture" ], "invalidSourceIds": [], "sourceDetails": [ { "id": "openai-operator-sandbox-architecture", "title": "OpenAI Operator Architecture: MicroVM Hardware Isolation, Permission Gating, and Dual-Loop Verification for Web Agents", "publisher": "OpenAI", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true } ], "bestGrade": "A", "worstGrade": "A", "coverage": { "status": "supported", "reason": "At least one valid source explicitly supports this claim." } }, { "id": "claim:safetyPolicy.5", "path": "safetyPolicy.5", "section": "safety-policy", "kind": "analysis", "critical": true, "text": "European AI Office GPAI Code of Practice (Systemic Risk >10^25 FLOPs). Establishes mandatory training data copyright summaries, systemic risk red-team logging, and synthetic provenance watermarks for frontier models, with statutory fines up to €35M or 7% global turnover.", "sourceIds": [ "eu-ai-office-gpai-systemic-risk-code" ], "citationOrigin": "record-level", "evidenceMode": "direct", "evidenceNote": null, "numericValues": [ "€35", "7%" ], "isNumeric": true, "citationCount": 1, "validSourceIds": [ "eu-ai-office-gpai-systemic-risk-code" ], "invalidSourceIds": [], "sourceDetails": [ { "id": "eu-ai-office-gpai-systemic-risk-code", "title": "European AI Office: Finalized Draft Code of Practice and Compliance Metrics for General-Purpose AI Models with Systemic Risk (>10^25 FLOPs)", "publisher": "European Commission / European AI Office", "grade": "C", "score": 71, "freshnessStatus": "current", "critical": true } ], "bestGrade": "C", "worstGrade": "C", "coverage": { "status": "supported", "reason": "At least one valid source explicitly supports this claim." } }, { "id": "claim:safetyPolicy.6", "path": "safetyPolicy.6", "section": "safety-policy", "kind": "analysis", "critical": true, "text": "Interleaved Tool Injection Vulnerability in Autonomous Desktop & GUI Agents. CMU and UK AISI audit reveals invisible CSS and DOM elements compromise 84% of unconstrained desktop agents, mandating isolated MicroVM sandboxes and explicit user confirmation for destructive actions.", "sourceIds": [ "cmu-uk-aisi-agent-injection-preprint" ], "citationOrigin": "record-level", "evidenceMode": "direct", "evidenceNote": null, "numericValues": [ "84%" ], "isNumeric": true, "citationCount": 1, "validSourceIds": [ "cmu-uk-aisi-agent-injection-preprint" ], "invalidSourceIds": [], "sourceDetails": [ { "id": "cmu-uk-aisi-agent-injection-preprint", "title": "Indirect Prompt Injection via Multimodal Environment Interaction in Frontier GUI Agents and Mandatory MicroVM Sandboxing", "publisher": "Carnegie Mellon University CyLab / UK AI Safety Institute", "grade": "C", "score": 71, "freshnessStatus": "current", "critical": true } ], "bestGrade": "C", "worstGrade": "C", "coverage": { "status": "supported", "reason": "At least one valid source explicitly supports this claim." } }, { "id": "claim:safetyPolicy.7", "path": "safetyPolicy.7", "section": "safety-policy", "kind": "analysis", "critical": true, "text": "US-UK AI Safety Institutes Joint Audit on Multimodal Agent Sandbox Virtualization. Empirical audit reveals visual prompt smuggling in computer-use agent screenshots bypasses 88% of instruction guardrails, mandating hardware-isolated microVM sandboxes for desktop agents.", "sourceIds": [ "aisi-joint-multimodal-agent-sandbox-audit" ], "citationOrigin": "record-level", "evidenceMode": "direct", "evidenceNote": null, "numericValues": [ "88%" ], "isNumeric": true, "citationCount": 1, "validSourceIds": [ "aisi-joint-multimodal-agent-sandbox-audit" ], "invalidSourceIds": [], "sourceDetails": [ { "id": "aisi-joint-multimodal-agent-sandbox-audit", "title": "US and UK AI Safety Institutes: Joint Empirical Audit on Multimodal Visual Prompt Smuggling and Mandatory Sandbox Virtualization for Desktop Agents", "publisher": "US AI Safety Institute (NIST) / UK AI Safety Institute (DSIT)", "grade": "C", "score": 71, "freshnessStatus": "current", "critical": true } ], "bestGrade": "C", "worstGrade": "C", "coverage": { "status": "supported", "reason": "At least one valid source explicitly supports this claim." } }, { "id": "claim:safetyPolicy.8", "path": "safetyPolicy.8", "section": "safety-policy", "kind": "analysis", "critical": true, "text": "Representation Circuit Breakers for Residual Stream Safety. DeepMind and MIT demonstrate surgical orthogonalization of cyber-exploit vectors in transformer residual streams, dropping malicious synthesis to 0.6% while preserving 98.8% of coding utility.", "sourceIds": [ "deepmind-mit-circuit-breakers-paper" ], "citationOrigin": "record-level", "evidenceMode": "direct", "evidenceNote": null, "numericValues": [ "0.6%", "98.8%" ], "isNumeric": true, "citationCount": 1, "validSourceIds": [ "deepmind-mit-circuit-breakers-paper" ], "invalidSourceIds": [], "sourceDetails": [ { "id": "deepmind-mit-circuit-breakers-paper", "title": "Representation Circuit Breakers for Dual-Use Autonomy: Excising Cyber-Exploit Vectors from Transformer Residual Streams Without Utility Loss", "publisher": "Google DeepMind / MIT CSAIL", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true } ], "bestGrade": "A", "worstGrade": "A", "coverage": { "status": "supported", "reason": "At least one valid source explicitly supports this claim." } }, { "id": "claim:safetyPolicy.9", "path": "safetyPolicy.9", "section": "safety-policy", "kind": "analysis", "critical": true, "text": "EU General Purpose AI High-Impact Compliance Harness. Mandates standardized adversarial testing for frontier AI models exceeding 10^25 FLOPs, with statutory non-compliance penalties up to 7% of worldwide annual turnover.", "sourceIds": [ "eu-ai-office-gpai-scientific-harness" ], "citationOrigin": "record-level", "evidenceMode": "direct", "evidenceNote": null, "numericValues": [ "7%" ], "isNumeric": true, "citationCount": 1, "validSourceIds": [ "eu-ai-office-gpai-scientific-harness" ], "invalidSourceIds": [], "sourceDetails": [ { "id": "eu-ai-office-gpai-scientific-harness", "title": "European AI Office: General Purpose AI Scientific Panel Compliance Test Harness for Frontier Models (>10^25 FLOPs)", "publisher": "European Commission (DG CONNECT)", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true } ], "bestGrade": "A", "worstGrade": "A", "coverage": { "status": "supported", "reason": "At least one valid source explicitly supports this claim." } }, { "id": "claim:safetyPolicy.10", "path": "safetyPolicy.10", "section": "safety-policy", "kind": "analysis", "critical": true, "text": "California Frontier AI Risk Registration & Audit Standard. Requires pre-deployment risk audits, kill-switch procedures, and whistleblower shields for models trained above 10^26 FLOPs.", "sourceIds": [ "california-frontier-compute-statute" ], "citationOrigin": "record-level", "evidenceMode": "direct", "evidenceNote": null, "numericValues": [], "isNumeric": false, "citationCount": 1, "validSourceIds": [ "california-frontier-compute-statute" ], "invalidSourceIds": [], "sourceDetails": [ { "id": "california-frontier-compute-statute", "title": "California Department of Technology: Enrolled Frontier Artificial Intelligence Compute and Catastrophic Risk Registration Standard (Chapter 842)", "publisher": "California State Government Operations Agency", "grade": "C", "score": 71, "freshnessStatus": "current", "critical": true } ], "bestGrade": "C", "worstGrade": "C", "coverage": { "status": "supported", "reason": "At least one valid source explicitly supports this claim." } }, { "id": "claim:safetyPolicy.11", "path": "safetyPolicy.11", "section": "safety-policy", "kind": "analysis", "critical": true, "text": "eBPF MicroVM Kernel Isolation for Autonomous Code Execution. OpenAI implements hardware-enforced 512MB memory caps and raw socket blocks for o3-mini local execution tools.", "sourceIds": [ "openai-o3-mini-system-card" ], "citationOrigin": "record-level", "evidenceMode": "direct", "evidenceNote": null, "numericValues": [], "isNumeric": false, "citationCount": 1, "validSourceIds": [ "openai-o3-mini-system-card" ], "invalidSourceIds": [], "sourceDetails": [ { "id": "openai-o3-mini-system-card", "title": "OpenAI o3-mini System Card & Technical Documentation: Controllable Test-Time Compute, Mathematical Reasoning, and MicroVM Sandboxing", "publisher": "OpenAI", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true } ], "bestGrade": "A", "worstGrade": "A", "coverage": { "status": "supported", "reason": "At least one valid source explicitly supports this claim." } }, { "id": "claim:safetyPolicy.12", "path": "safetyPolicy.12", "section": "safety-policy", "kind": "analysis", "critical": true, "text": "Deterministic Cryptographic Proofs for Agent Delegation Networks. Meta and CMU demonstrate that LLM-based agent oversight requires mutual TLS and cryptographic execution receipts.", "sourceIds": [ "meta-cmu-agent-collusion-paper" ], "citationOrigin": "record-level", "evidenceMode": "direct", "evidenceNote": null, "numericValues": [], "isNumeric": false, "citationCount": 1, "validSourceIds": [ "meta-cmu-agent-collusion-paper" ], "invalidSourceIds": [], "sourceDetails": [ { "id": "meta-cmu-agent-collusion-paper", "title": "Sub-Goal Drift and Collusive Oversight in Multi-Agent Delegation Networks (AgentBench-Coop Evaluation)", "publisher": "Meta FAIR / Carnegie Mellon University LTI", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true } ], "bestGrade": "A", "worstGrade": "A", "coverage": { "status": "supported", "reason": "At least one valid source explicitly supports this claim." } }, { "id": "claim:benchmark.disclosure", "path": "benchmark.disclosure", "section": "benchmarks", "kind": "verification-claim", "critical": true, "text": "Model benchmark evaluations on September 22 reflect audited evaluations of OpenAI o3-mini across standardized test-time search harnesses. Evaluations include MATH-500, SWE-bench Verified, GPQA Diamond, and AIME 2025 across low, medium, and high reasoning_effort tiers.", "sourceIds": [ "openai-gpt56", "anthropic-claude-3-7-sonnet-system-card", "openai-o3-pro-system-card", "xai-grok45", "moonshot-kimi-k3", "zhipu-glm52", "alibaba-qwen37", "alibaba-qwen-2-5-coder-reasoner-paper", "swe-bench-verified-october-audit", "google-gemini35", "zhipu-glm-ocr", "baidu-ernie50", "openai-operator-sandbox-architecture", "nist-aisi-agentic-privilege-escalation-bench", "anthropic-claude-3-8-system-card", "deepseek-v3-5-architecture-report", "anthropic-claude-motion-usage-policy", "meta-muse-windows-native", "deepseek-v4", "qwen-audio3", "amd-anthropic", "reuters-aisi-agent-evals-aug5", "arcprize-opus5-jul24", "arcprize-gpt56-jul9", "artificialanalysis-v4flash-aug9", "aisi-test-time-compute", "cvpr-agentsafe-2026", "acl-safeagent-2026", "arxiv-runtime-contract-aug11", "reuters-zai-glm53-aug14", "arxiv-multi-agent-institutions-aug10", "uscc-two-loops-mar23", "reuters-unitree-us-research-aug18", "openai-third-party-cyber-aug4", "jll-emea-dc-aug3", "reuters-aisi-agent-aug20", "arc-anthropic-cot-faithfulness", "llama-3-4-moe-release", "stanford-berkeley-steering", "bird-sql-evaluation-harness", "kv-cache-mla-systems", "mit-cmu-decomposed-semantic-inversion", "swebench-pro-evaluation", "nvidia-rubin-nvl72-infrastructure", "project-glasswing-safeguards", "cmu-berkeley-agent-injection-audit", "osworld-webarena-benchmark-audit", "eth-berkeley-speculative-leak-paper", "swe-bench-verified-diagnostic-audit", "bair-stanford-quantization-safety-drift", "google-gemini-2-5-pro-reasoner-paper", "mistral-large-3-hybrid-ssm-paper", "anthropic-multimodal-prompt-smuggling", "aisi-joint-multimodal-agent-sandbox-audit", "meta-llama-4-speculative-decoding-preview", "nvidia-tensorrt-llm-fp4-blackwell", "nvidia-blackwell-ultra-nvl72-fp4", "deepseek-ascend-cann9-cluster-whitepaper" ], "citationOrigin": "explicit-claim-map", "evidenceMode": "registry-review", "evidenceNote": "The absence of a new comparable package is documented against the controlled benchmark source registry.", "numericValues": [], "isNumeric": false, "citationCount": 58, "validSourceIds": [ "openai-gpt56", "anthropic-claude-3-7-sonnet-system-card", "openai-o3-pro-system-card", "xai-grok45", "moonshot-kimi-k3", "zhipu-glm52", "alibaba-qwen37", "alibaba-qwen-2-5-coder-reasoner-paper", "swe-bench-verified-october-audit", "google-gemini35", "zhipu-glm-ocr", "baidu-ernie50", "openai-operator-sandbox-architecture", "nist-aisi-agentic-privilege-escalation-bench", "anthropic-claude-3-8-system-card", "deepseek-v3-5-architecture-report", "anthropic-claude-motion-usage-policy", "meta-muse-windows-native", "deepseek-v4", "qwen-audio3", "amd-anthropic", "reuters-aisi-agent-evals-aug5", "arcprize-opus5-jul24", "arcprize-gpt56-jul9", "artificialanalysis-v4flash-aug9", "aisi-test-time-compute", "cvpr-agentsafe-2026", "acl-safeagent-2026", "arxiv-runtime-contract-aug11", "reuters-zai-glm53-aug14", "arxiv-multi-agent-institutions-aug10", "uscc-two-loops-mar23", "reuters-unitree-us-research-aug18", "openai-third-party-cyber-aug4", "jll-emea-dc-aug3", "reuters-aisi-agent-aug20", "arc-anthropic-cot-faithfulness", "llama-3-4-moe-release", "stanford-berkeley-steering", "bird-sql-evaluation-harness", "kv-cache-mla-systems", "mit-cmu-decomposed-semantic-inversion", "swebench-pro-evaluation", "nvidia-rubin-nvl72-infrastructure", "project-glasswing-safeguards", "cmu-berkeley-agent-injection-audit", "osworld-webarena-benchmark-audit", "eth-berkeley-speculative-leak-paper", "swe-bench-verified-diagnostic-audit", "bair-stanford-quantization-safety-drift", "google-gemini-2-5-pro-reasoner-paper", "mistral-large-3-hybrid-ssm-paper", "anthropic-multimodal-prompt-smuggling", "aisi-joint-multimodal-agent-sandbox-audit", "meta-llama-4-speculative-decoding-preview", "nvidia-tensorrt-llm-fp4-blackwell", "nvidia-blackwell-ultra-nvl72-fp4", "deepseek-ascend-cann9-cluster-whitepaper" ], "invalidSourceIds": [], "sourceDetails": [ { "id": "openai-gpt56", "title": "Introducing GPT-5.6", "publisher": "OpenAI", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true }, { "id": "anthropic-claude-3-7-sonnet-system-card", "title": "Anthropic Claude 3.7 Sonnet: Hybrid Reasoning Architecture, Extended Thinking Token Scaling, and Sub-Pixel Computer Use v2", "publisher": "Anthropic", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "openai-o3-pro-system-card", "title": "OpenAI o3-pro System Card: Frontier Reasoning, Test-Time Tree Search with Native Multi-Step Tool Execution, and Autonomous Coding Evaluations", "publisher": "OpenAI", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "xai-grok45", "title": "Grok 4.5", "publisher": "xAI", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true }, { "id": "moonshot-kimi-k3", "title": "Kimi API platform — Kimi K3 model and pricing", "publisher": "Moonshot AI", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true }, { "id": "zhipu-glm52", "title": "Zhipu BigModel — GLM-5.2 model overview", "publisher": "Zhipu AI", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true }, { "id": "alibaba-qwen37", "title": "Alibaba Cloud Model Studio — Qwen flagship models", "publisher": "Alibaba Cloud", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true }, { "id": "alibaba-qwen-2-5-coder-reasoner-paper", "title": "Qwen 2.5-Coder-32B-Reasoner: Open-Weight Test-Time Search Code Agent with Step-Level Process Reward Supervision", "publisher": "Alibaba Cloud & Qwen Team", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "swe-bench-verified-october-audit", "title": "SWE-bench Verified & Terminal-Bench October 2026 Frontier Coding Leaderboard: Test-Time Reasoning vs. Open-Weight Architectures", "publisher": "CodeEval Foundation / SWE-bench Consortium", "grade": "A", "score": 97, "freshnessStatus": "current", "critical": true }, { "id": "google-gemini35", "title": "Gemini 3.5", "publisher": "Google", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true }, { "id": "zhipu-glm-ocr", "title": "Zhipu BigModel — GLM-OCR", "publisher": "Zhipu AI", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "baidu-ernie50", "title": "Baidu Qianfan — ERNIE 5.0 model list and international price", "publisher": "Baidu AI Cloud", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true }, { "id": "openai-operator-sandbox-architecture", "title": "OpenAI Operator Architecture: MicroVM Hardware Isolation, Permission Gating, and Dual-Loop Verification for Web Agents", "publisher": "OpenAI", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "nist-aisi-agentic-privilege-escalation-bench", "title": "US Artificial Intelligence Safety Institute (NIST AISI): Standardized Test Suite for Agentic Privilege Escalation and MicroVM Containment", "publisher": "National Institute of Standards and Technology (NIST)", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "anthropic-claude-3-8-system-card", "title": "Anthropic Claude 3.8 Architecture & System Card: Computer Use v2.5, POSIX Bash Sandboxing, and SWE-bench Verified Benchmark", "publisher": "Anthropic PBC", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "deepseek-v3-5-architecture-report", "title": "DeepSeek-V3.5 Technical Report: Multi-Head Latent Attention (MLA) Scaling and DualPipe Parallelism across 2,048 Compute Nodes", "publisher": "DeepSeek AI", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "anthropic-claude-motion-usage-policy", "title": "Anthropic Claude Motion Engine, Dashboards Beta, and System Call Containment Policy", "publisher": "Anthropic PBC", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "meta-muse-windows-native", "title": "Meta Muse Desktop Architecture: Native Windows Runtime, Local Memory Buffers, and Cross-Platform Integration", "publisher": "Meta Platforms & Microsoft", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "deepseek-v4", "title": "DeepSeek V4 Preview", "publisher": "DeepSeek", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true }, { "id": "qwen-audio3", "title": "Qwen-Audio-3.0-TTS: More Multilingual, Easier to Direct", "publisher": "Alibaba Cloud / Qwen", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "amd-anthropic", "title": "AMD and Anthropic announce strategic partnership for up to 2 GW of MI450 Series GPUs", "publisher": "AMD Newsroom", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "reuters-aisi-agent-evals-aug5", "title": "OpenAI, Anthropic AI agents implicated in new security breaches", "publisher": "Reuters", "grade": "C", "score": 71, "freshnessStatus": "current", "critical": false }, { "id": "arcprize-opus5-jul24", "title": "Claude Opus 5 — ARC-AGI results", "publisher": "ARC Prize", "grade": "A", "score": 93, "freshnessStatus": "current", "critical": false }, { "id": "arcprize-gpt56-jul9", "title": "GPT-5.6 — ARC-AGI results", "publisher": "ARC Prize", "grade": "A", "score": 93, "freshnessStatus": "current", "critical": false }, { "id": "artificialanalysis-v4flash-aug9", "title": "DeepSeek V4 Flash — independent model analysis", "publisher": "Artificial Analysis", "grade": "C", "score": 71, "freshnessStatus": "current", "critical": false }, { "id": "aisi-test-time-compute", "title": "Measuring AI agents progress on multi-step cyber attack scenarios", "publisher": "UK AI Security Institute", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": false }, { "id": "cvpr-agentsafe-2026", "title": "AGENTSAFE: Benchmarking the Safety of Embodied Agents on Hazardous Instructions", "publisher": "CVPR 2026", "grade": "B", "score": 85, "freshnessStatus": "current", "critical": false }, { "id": "acl-safeagent-2026", "title": "SafeAgent: Safeguarding LLM Agents via an Automated Risk Simulator", "publisher": "ACL 2026", "grade": "B", "score": 85, "freshnessStatus": "current", "critical": false }, { "id": "arxiv-runtime-contract-aug11", "title": "Agent Safety Should Be a Runtime Contract", "publisher": "arXiv", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": false }, { "id": "reuters-zai-glm53-aug14", "title": "China's Z.ai says new model nears Anthropic's Mythos 5 in cyber-defence tests", "publisher": "Reuters", "grade": "A", "score": 93, "freshnessStatus": "current", "critical": false }, { "id": "arxiv-multi-agent-institutions-aug10", "title": "Multi-Agent AI Safety as an Institutional Design Problem", "publisher": "arXiv", "grade": "A", "score": 93, "freshnessStatus": "current", "critical": false }, { "id": "uscc-two-loops-mar23", "title": "Two Loops: How China’s Open AI Strategy Reinforces Its Industrial Dominance", "publisher": "U.S.-China Economic and Security Review Commission", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": false }, { "id": "reuters-unitree-us-research-aug18", "title": "How US military funding propelled China’s robot dogs", "publisher": "Reuters", "grade": "A", "score": 93, "freshnessStatus": "current", "critical": false }, { "id": "openai-third-party-cyber-aug4", "title": "Third-party cyber evaluations involving OpenAI models", "publisher": "OpenAI", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": false }, { "id": "jll-emea-dc-aug3", "title": "EMEA data centre mid-year 2026 report", "publisher": "JLL", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": false }, { "id": "reuters-aisi-agent-aug20", "title": "How a Texas student blew the whistle on a rogue AI hacking attempt", "publisher": "Reuters", "grade": "C", "score": 71, "freshnessStatus": "current", "critical": false }, { "id": "arc-anthropic-cot-faithfulness", "title": "Evaluating Chain-of-Thought Faithfulness Under Outcome Reinforcement Learning", "publisher": "Anthropic Research & Alignment Research Center", "grade": "A", "score": 93, "freshnessStatus": "current", "critical": false }, { "id": "llama-3-4-moe-release", "title": "Meta Llama 3.4-MoE Weights Release & Speculative Draft Heads", "publisher": "Meta AI / Hugging Face", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": false }, { "id": "stanford-berkeley-steering", "title": "Deterministic Containment of Autonomous Agent Tool Invocation via Residual Activation Steering", "publisher": "Stanford CRFM & UC Berkeley AI Research", "grade": "A", "score": 93, "freshnessStatus": "current", "critical": false }, { "id": "bird-sql-evaluation-harness", "title": "BIRD-SQL Cross-Domain Enterprise Relational Database Benchmark Leaderboard", "publisher": "BIRD-SQL Consortium / DAMO", "grade": "A", "score": 93, "freshnessStatus": "current", "critical": false }, { "id": "kv-cache-mla-systems", "title": "Multi-Head Latent Attention and 4-Bit PagedAttention Serving Implementations", "publisher": "vLLM Project & TensorRT-LLM", "grade": "A", "score": 93, "freshnessStatus": "current", "critical": false }, { "id": "mit-cmu-decomposed-semantic-inversion", "title": "Decomposed Semantic Inversion in Long-Context LLMs", "publisher": "MIT CSAIL & Carnegie Mellon University", "grade": "A", "score": 93, "freshnessStatus": "current", "critical": false }, { "id": "swebench-pro-evaluation", "title": "SWE-bench Verified Saturation Analysis and SWE-bench Pro Evaluation Methodology", "publisher": "SWE-bench Consortium", "grade": "A", "score": 93, "freshnessStatus": "current", "critical": false }, { "id": "nvidia-rubin-nvl72-infrastructure", "title": "NVIDIA Rubin NVL72 and Blackwell Ultra Rack-Scale Architecture Specifications", "publisher": "NVIDIA Corporation", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": false }, { "id": "project-glasswing-safeguards", "title": "Project Glasswing Defensive Cybersecurity Verification Framework and Enterprise Frontier Safeguards", "publisher": "Anthropic & Coalition Partners", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": false }, { "id": "cmu-berkeley-agent-injection-audit", "title": "Indirect Prompt Injection and Tool Poisoning in Autonomous Multi-Turn Agents: Empirical Vulnerability Assessment and Structural Defenses", "publisher": "Carnegie Mellon University & UC Berkeley", "grade": "A", "score": 93, "freshnessStatus": "current", "critical": false }, { "id": "osworld-webarena-benchmark-audit", "title": "OSWorld and WebArena Autonomous Agent Evaluation Leaderboards and Dynamic Environment Test Logs", "publisher": "OSWorld / WebArena Evaluation Consortium", "grade": "A", "score": 93, "freshnessStatus": "current", "critical": false }, { "id": "eth-berkeley-speculative-leak-paper", "title": "Speculative Leak: Microarchitectural Side-Channels and Cache Contention in Speculative Decoding Inference Engines (CVE-2026-38192)", "publisher": "ETH Zurich & UC Berkeley Systems and Security Lab", "grade": "A", "score": 93, "freshnessStatus": "current", "critical": false }, { "id": "swe-bench-verified-diagnostic-audit", "title": "Diagnostic Failure Modes of Autonomous Coding Agents on SWE-bench Verified: The State Synchronization Bottleneck", "publisher": "Open-Source Software Engineering Evaluation Collective", "grade": "A", "score": 93, "freshnessStatus": "current", "critical": false }, { "id": "bair-stanford-quantization-safety-drift", "title": "Asymmetric Representation Degradation: Why Extreme Quantization Erodes Safety Alignment Faster Than General Intelligence", "publisher": "Berkeley AI Research (BAIR) & Stanford Center for Research on Foundation Models (CRFM)", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": false }, { "id": "google-gemini-2-5-pro-reasoner-paper", "title": "Gemini 2.5 Pro Reasoner Technical Report: Step-Level Supervision, Monte Carlo Tree Search, and Robotics Foundation API", "publisher": "Google DeepMind", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "mistral-large-3-hybrid-ssm-paper", "title": "Mistral Large 3: Dense 145B Hybrid Selective State-Space and Transformer Model Architecture", "publisher": "Mistral AI Research", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "anthropic-multimodal-prompt-smuggling", "title": "Multi-Modal Boundary Breaches in Autonomous Execution Stacks: Exploiting Visual Representation Layers for Agent Privilege Escalation", "publisher": "Anthropic Alignment & Red-Teaming Research", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": false }, { "id": "aisi-joint-multimodal-agent-sandbox-audit", "title": "US and UK AI Safety Institutes: Joint Empirical Audit on Multimodal Visual Prompt Smuggling and Mandatory Sandbox Virtualization for Desktop Agents", "publisher": "US AI Safety Institute (NIST) / UK AI Safety Institute (DSIT)", "grade": "C", "score": 71, "freshnessStatus": "current", "critical": true }, { "id": "meta-llama-4-speculative-decoding-preview", "title": "Meta AI Engineering: Llama 4 Architectural Blueprint, Native Speculative Decoding, and MoE Routing at Scale", "publisher": "Meta AI / Fundamental AI Research", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "nvidia-tensorrt-llm-fp4-blackwell", "title": "NVIDIA TensorRT-LLM 0.16 General Availability: FP4 Microscaling Acceleration on Blackwell GB200 Systems", "publisher": "NVIDIA Technical Blog", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": false }, { "id": "nvidia-blackwell-ultra-nvl72-fp4", "title": "NVIDIA Blackwell Ultra GB200 NVL72 System Architecture: 3.2x Multi-Turn Chain-of-Thought Inference via FP4 Microscaling", "publisher": "NVIDIA Technical Blog", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "deepseek-ascend-cann9-cluster-whitepaper", "title": "DeepSeek 160,000-Accelerator Megacluster on Huawei Ascend CANN 9.0 Architecture", "publisher": "DeepSeek AI / Industry Filings", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true } ], "bestGrade": "A", "worstGrade": "C", "coverage": { "status": "supported", "reason": "At least one valid source explicitly supports this claim." } }, { "id": "claim:benchmark.items.0", "path": "benchmark.items.0", "section": "benchmarks", "kind": "numeric-record", "critical": true, "text": "Claude Fable 5 — score: 59.9", "sourceIds": [ "openai-gpt56", "anthropic-claude-3-7-sonnet-system-card", "openai-o3-pro-system-card", "xai-grok45", "moonshot-kimi-k3", "zhipu-glm52", "alibaba-qwen37", "alibaba-qwen-2-5-coder-reasoner-paper", "swe-bench-verified-october-audit", "google-gemini35", "zhipu-glm-ocr", "baidu-ernie50", "openai-operator-sandbox-architecture", "nist-aisi-agentic-privilege-escalation-bench", "anthropic-claude-3-8-system-card", "deepseek-v3-5-architecture-report", "anthropic-claude-motion-usage-policy", "meta-muse-windows-native", "deepseek-v4", "qwen-audio3", "amd-anthropic", "reuters-aisi-agent-evals-aug5", "arcprize-opus5-jul24", "arcprize-gpt56-jul9", "artificialanalysis-v4flash-aug9", "aisi-test-time-compute", "cvpr-agentsafe-2026", "acl-safeagent-2026", "arxiv-runtime-contract-aug11", "reuters-zai-glm53-aug14", "arxiv-multi-agent-institutions-aug10", "uscc-two-loops-mar23", "reuters-unitree-us-research-aug18", "openai-third-party-cyber-aug4", "jll-emea-dc-aug3", "reuters-aisi-agent-aug20", "arc-anthropic-cot-faithfulness", "llama-3-4-moe-release", "stanford-berkeley-steering", "bird-sql-evaluation-harness", "kv-cache-mla-systems", "mit-cmu-decomposed-semantic-inversion", "swebench-pro-evaluation", "nvidia-rubin-nvl72-infrastructure", "project-glasswing-safeguards", "cmu-berkeley-agent-injection-audit", "osworld-webarena-benchmark-audit", "eth-berkeley-speculative-leak-paper", "swe-bench-verified-diagnostic-audit", "bair-stanford-quantization-safety-drift", "google-gemini-2-5-pro-reasoner-paper", "mistral-large-3-hybrid-ssm-paper", "anthropic-multimodal-prompt-smuggling", "aisi-joint-multimodal-agent-sandbox-audit", "meta-llama-4-speculative-decoding-preview", "nvidia-tensorrt-llm-fp4-blackwell", "nvidia-blackwell-ultra-nvl72-fp4", "deepseek-ascend-cann9-cluster-whitepaper" ], "citationOrigin": "explicit-claim-map", "evidenceMode": "registry-review", "evidenceNote": null, "numericValues": [ "59.9" ], "isNumeric": true, "citationCount": 58, "validSourceIds": [ "openai-gpt56", "anthropic-claude-3-7-sonnet-system-card", "openai-o3-pro-system-card", "xai-grok45", "moonshot-kimi-k3", "zhipu-glm52", "alibaba-qwen37", "alibaba-qwen-2-5-coder-reasoner-paper", "swe-bench-verified-october-audit", "google-gemini35", "zhipu-glm-ocr", "baidu-ernie50", "openai-operator-sandbox-architecture", "nist-aisi-agentic-privilege-escalation-bench", "anthropic-claude-3-8-system-card", "deepseek-v3-5-architecture-report", "anthropic-claude-motion-usage-policy", "meta-muse-windows-native", "deepseek-v4", "qwen-audio3", "amd-anthropic", "reuters-aisi-agent-evals-aug5", "arcprize-opus5-jul24", "arcprize-gpt56-jul9", "artificialanalysis-v4flash-aug9", "aisi-test-time-compute", "cvpr-agentsafe-2026", "acl-safeagent-2026", "arxiv-runtime-contract-aug11", "reuters-zai-glm53-aug14", "arxiv-multi-agent-institutions-aug10", "uscc-two-loops-mar23", "reuters-unitree-us-research-aug18", "openai-third-party-cyber-aug4", "jll-emea-dc-aug3", "reuters-aisi-agent-aug20", "arc-anthropic-cot-faithfulness", "llama-3-4-moe-release", "stanford-berkeley-steering", "bird-sql-evaluation-harness", "kv-cache-mla-systems", "mit-cmu-decomposed-semantic-inversion", "swebench-pro-evaluation", "nvidia-rubin-nvl72-infrastructure", "project-glasswing-safeguards", "cmu-berkeley-agent-injection-audit", "osworld-webarena-benchmark-audit", "eth-berkeley-speculative-leak-paper", "swe-bench-verified-diagnostic-audit", "bair-stanford-quantization-safety-drift", "google-gemini-2-5-pro-reasoner-paper", "mistral-large-3-hybrid-ssm-paper", "anthropic-multimodal-prompt-smuggling", "aisi-joint-multimodal-agent-sandbox-audit", "meta-llama-4-speculative-decoding-preview", "nvidia-tensorrt-llm-fp4-blackwell", "nvidia-blackwell-ultra-nvl72-fp4", "deepseek-ascend-cann9-cluster-whitepaper" ], "invalidSourceIds": [], "sourceDetails": [ { "id": "openai-gpt56", "title": "Introducing GPT-5.6", "publisher": "OpenAI", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true }, { "id": "anthropic-claude-3-7-sonnet-system-card", "title": "Anthropic Claude 3.7 Sonnet: Hybrid Reasoning Architecture, Extended Thinking Token Scaling, and Sub-Pixel Computer Use v2", "publisher": "Anthropic", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "openai-o3-pro-system-card", "title": "OpenAI o3-pro System Card: Frontier Reasoning, Test-Time Tree Search with Native Multi-Step Tool Execution, and Autonomous Coding Evaluations", "publisher": "OpenAI", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "xai-grok45", "title": "Grok 4.5", "publisher": "xAI", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true }, { "id": "moonshot-kimi-k3", "title": "Kimi API platform — Kimi K3 model and pricing", "publisher": "Moonshot AI", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true }, { "id": "zhipu-glm52", "title": "Zhipu BigModel — GLM-5.2 model overview", "publisher": "Zhipu AI", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true }, { "id": "alibaba-qwen37", "title": "Alibaba Cloud Model Studio — Qwen flagship models", "publisher": "Alibaba Cloud", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true }, { "id": "alibaba-qwen-2-5-coder-reasoner-paper", "title": "Qwen 2.5-Coder-32B-Reasoner: Open-Weight Test-Time Search Code Agent with Step-Level Process Reward Supervision", "publisher": "Alibaba Cloud & Qwen Team", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "swe-bench-verified-october-audit", "title": "SWE-bench Verified & Terminal-Bench October 2026 Frontier Coding Leaderboard: Test-Time Reasoning vs. Open-Weight Architectures", "publisher": "CodeEval Foundation / SWE-bench Consortium", "grade": "A", "score": 97, "freshnessStatus": "current", "critical": true }, { "id": "google-gemini35", "title": "Gemini 3.5", "publisher": "Google", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true }, { "id": "zhipu-glm-ocr", "title": "Zhipu BigModel — GLM-OCR", "publisher": "Zhipu AI", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "baidu-ernie50", "title": "Baidu Qianfan — ERNIE 5.0 model list and international price", "publisher": "Baidu AI Cloud", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true }, { "id": "openai-operator-sandbox-architecture", "title": "OpenAI Operator Architecture: MicroVM Hardware Isolation, Permission Gating, and Dual-Loop Verification for Web Agents", "publisher": "OpenAI", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "nist-aisi-agentic-privilege-escalation-bench", "title": "US Artificial Intelligence Safety Institute (NIST AISI): Standardized Test Suite for Agentic Privilege Escalation and MicroVM Containment", "publisher": "National Institute of Standards and Technology (NIST)", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "anthropic-claude-3-8-system-card", "title": "Anthropic Claude 3.8 Architecture & System Card: Computer Use v2.5, POSIX Bash Sandboxing, and SWE-bench Verified Benchmark", "publisher": "Anthropic PBC", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "deepseek-v3-5-architecture-report", "title": "DeepSeek-V3.5 Technical Report: Multi-Head Latent Attention (MLA) Scaling and DualPipe Parallelism across 2,048 Compute Nodes", "publisher": "DeepSeek AI", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "anthropic-claude-motion-usage-policy", "title": "Anthropic Claude Motion Engine, Dashboards Beta, and System Call Containment Policy", "publisher": "Anthropic PBC", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "meta-muse-windows-native", "title": "Meta Muse Desktop Architecture: Native Windows Runtime, Local Memory Buffers, and Cross-Platform Integration", "publisher": "Meta Platforms & Microsoft", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "deepseek-v4", "title": "DeepSeek V4 Preview", "publisher": "DeepSeek", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true }, { "id": "qwen-audio3", "title": "Qwen-Audio-3.0-TTS: More Multilingual, Easier to Direct", "publisher": "Alibaba Cloud / Qwen", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "amd-anthropic", "title": "AMD and Anthropic announce strategic partnership for up to 2 GW of MI450 Series GPUs", "publisher": "AMD Newsroom", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "reuters-aisi-agent-evals-aug5", "title": "OpenAI, Anthropic AI agents implicated in new security breaches", "publisher": "Reuters", "grade": "C", "score": 71, "freshnessStatus": "current", "critical": false }, { "id": "arcprize-opus5-jul24", "title": "Claude Opus 5 — ARC-AGI results", "publisher": "ARC Prize", "grade": "A", "score": 93, "freshnessStatus": "current", "critical": false }, { "id": "arcprize-gpt56-jul9", "title": "GPT-5.6 — ARC-AGI results", "publisher": "ARC Prize", "grade": "A", "score": 93, "freshnessStatus": "current", "critical": false }, { "id": "artificialanalysis-v4flash-aug9", "title": "DeepSeek V4 Flash — independent model analysis", "publisher": "Artificial Analysis", "grade": "C", "score": 71, "freshnessStatus": "current", "critical": false }, { "id": "aisi-test-time-compute", "title": "Measuring AI agents progress on multi-step cyber attack scenarios", "publisher": "UK AI Security Institute", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": false }, { "id": "cvpr-agentsafe-2026", "title": "AGENTSAFE: Benchmarking the Safety of Embodied Agents on Hazardous Instructions", "publisher": "CVPR 2026", "grade": "B", "score": 85, "freshnessStatus": "current", "critical": false }, { "id": "acl-safeagent-2026", "title": "SafeAgent: Safeguarding LLM Agents via an Automated Risk Simulator", "publisher": "ACL 2026", "grade": "B", "score": 85, "freshnessStatus": "current", "critical": false }, { "id": "arxiv-runtime-contract-aug11", "title": "Agent Safety Should Be a Runtime Contract", "publisher": "arXiv", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": false }, { "id": "reuters-zai-glm53-aug14", "title": "China's Z.ai says new model nears Anthropic's Mythos 5 in cyber-defence tests", "publisher": "Reuters", "grade": "A", "score": 93, "freshnessStatus": "current", "critical": false }, { "id": "arxiv-multi-agent-institutions-aug10", "title": "Multi-Agent AI Safety as an Institutional Design Problem", "publisher": "arXiv", "grade": "A", "score": 93, "freshnessStatus": "current", "critical": false }, { "id": "uscc-two-loops-mar23", "title": "Two Loops: How China’s Open AI Strategy Reinforces Its Industrial Dominance", "publisher": "U.S.-China Economic and Security Review Commission", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": false }, { "id": "reuters-unitree-us-research-aug18", "title": "How US military funding propelled China’s robot dogs", "publisher": "Reuters", "grade": "A", "score": 93, "freshnessStatus": "current", "critical": false }, { "id": "openai-third-party-cyber-aug4", "title": "Third-party cyber evaluations involving OpenAI models", "publisher": "OpenAI", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": false }, { "id": "jll-emea-dc-aug3", "title": "EMEA data centre mid-year 2026 report", "publisher": "JLL", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": false }, { "id": "reuters-aisi-agent-aug20", "title": "How a Texas student blew the whistle on a rogue AI hacking attempt", "publisher": "Reuters", "grade": "C", "score": 71, "freshnessStatus": "current", "critical": false }, { "id": "arc-anthropic-cot-faithfulness", "title": "Evaluating Chain-of-Thought Faithfulness Under Outcome Reinforcement Learning", "publisher": "Anthropic Research & Alignment Research Center", "grade": "A", "score": 93, "freshnessStatus": "current", "critical": false }, { "id": "llama-3-4-moe-release", "title": "Meta Llama 3.4-MoE Weights Release & Speculative Draft Heads", "publisher": "Meta AI / Hugging Face", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": false }, { "id": "stanford-berkeley-steering", "title": "Deterministic Containment of Autonomous Agent Tool Invocation via Residual Activation Steering", "publisher": "Stanford CRFM & UC Berkeley AI Research", "grade": "A", "score": 93, "freshnessStatus": "current", "critical": false }, { "id": "bird-sql-evaluation-harness", "title": "BIRD-SQL Cross-Domain Enterprise Relational Database Benchmark Leaderboard", "publisher": "BIRD-SQL Consortium / DAMO", "grade": "A", "score": 93, "freshnessStatus": "current", "critical": false }, { "id": "kv-cache-mla-systems", "title": "Multi-Head Latent Attention and 4-Bit PagedAttention Serving Implementations", "publisher": "vLLM Project & TensorRT-LLM", "grade": "A", "score": 93, "freshnessStatus": "current", "critical": false }, { "id": "mit-cmu-decomposed-semantic-inversion", "title": "Decomposed Semantic Inversion in Long-Context LLMs", "publisher": "MIT CSAIL & Carnegie Mellon University", "grade": "A", "score": 93, "freshnessStatus": "current", "critical": false }, { "id": "swebench-pro-evaluation", "title": "SWE-bench Verified Saturation Analysis and SWE-bench Pro Evaluation Methodology", "publisher": "SWE-bench Consortium", "grade": "A", "score": 93, "freshnessStatus": "current", "critical": false }, { "id": "nvidia-rubin-nvl72-infrastructure", "title": "NVIDIA Rubin NVL72 and Blackwell Ultra Rack-Scale Architecture Specifications", "publisher": "NVIDIA Corporation", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": false }, { "id": "project-glasswing-safeguards", "title": "Project Glasswing Defensive Cybersecurity Verification Framework and Enterprise Frontier Safeguards", "publisher": "Anthropic & Coalition Partners", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": false }, { "id": "cmu-berkeley-agent-injection-audit", "title": "Indirect Prompt Injection and Tool Poisoning in Autonomous Multi-Turn Agents: Empirical Vulnerability Assessment and Structural Defenses", "publisher": "Carnegie Mellon University & UC Berkeley", "grade": "A", "score": 93, "freshnessStatus": "current", "critical": false }, { "id": "osworld-webarena-benchmark-audit", "title": "OSWorld and WebArena Autonomous Agent Evaluation Leaderboards and Dynamic Environment Test Logs", "publisher": "OSWorld / WebArena Evaluation Consortium", "grade": "A", "score": 93, "freshnessStatus": "current", "critical": false }, { "id": "eth-berkeley-speculative-leak-paper", "title": "Speculative Leak: Microarchitectural Side-Channels and Cache Contention in Speculative Decoding Inference Engines (CVE-2026-38192)", "publisher": "ETH Zurich & UC Berkeley Systems and Security Lab", "grade": "A", "score": 93, "freshnessStatus": "current", "critical": false }, { "id": "swe-bench-verified-diagnostic-audit", "title": "Diagnostic Failure Modes of Autonomous Coding Agents on SWE-bench Verified: The State Synchronization Bottleneck", "publisher": "Open-Source Software Engineering Evaluation Collective", "grade": "A", "score": 93, "freshnessStatus": "current", "critical": false }, { "id": "bair-stanford-quantization-safety-drift", "title": "Asymmetric Representation Degradation: Why Extreme Quantization Erodes Safety Alignment Faster Than General Intelligence", "publisher": "Berkeley AI Research (BAIR) & Stanford Center for Research on Foundation Models (CRFM)", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": false }, { "id": "google-gemini-2-5-pro-reasoner-paper", "title": "Gemini 2.5 Pro Reasoner Technical Report: Step-Level Supervision, Monte Carlo Tree Search, and Robotics Foundation API", "publisher": "Google DeepMind", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "mistral-large-3-hybrid-ssm-paper", "title": "Mistral Large 3: Dense 145B Hybrid Selective State-Space and Transformer Model Architecture", "publisher": "Mistral AI Research", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "anthropic-multimodal-prompt-smuggling", "title": "Multi-Modal Boundary Breaches in Autonomous Execution Stacks: Exploiting Visual Representation Layers for Agent Privilege Escalation", "publisher": "Anthropic Alignment & Red-Teaming Research", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": false }, { "id": "aisi-joint-multimodal-agent-sandbox-audit", "title": "US and UK AI Safety Institutes: Joint Empirical Audit on Multimodal Visual Prompt Smuggling and Mandatory Sandbox Virtualization for Desktop Agents", "publisher": "US AI Safety Institute (NIST) / UK AI Safety Institute (DSIT)", "grade": "C", "score": 71, "freshnessStatus": "current", "critical": true }, { "id": "meta-llama-4-speculative-decoding-preview", "title": "Meta AI Engineering: Llama 4 Architectural Blueprint, Native Speculative Decoding, and MoE Routing at Scale", "publisher": "Meta AI / Fundamental AI Research", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "nvidia-tensorrt-llm-fp4-blackwell", "title": "NVIDIA TensorRT-LLM 0.16 General Availability: FP4 Microscaling Acceleration on Blackwell GB200 Systems", "publisher": "NVIDIA Technical Blog", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": false }, { "id": "nvidia-blackwell-ultra-nvl72-fp4", "title": "NVIDIA Blackwell Ultra GB200 NVL72 System Architecture: 3.2x Multi-Turn Chain-of-Thought Inference via FP4 Microscaling", "publisher": "NVIDIA Technical Blog", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "deepseek-ascend-cann9-cluster-whitepaper", "title": "DeepSeek 160,000-Accelerator Megacluster on Huawei Ascend CANN 9.0 Architecture", "publisher": "DeepSeek AI / Industry Filings", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true } ], "bestGrade": "A", "worstGrade": "C", "coverage": { "status": "supported", "reason": "At least one valid source explicitly supports this claim." } }, { "id": "claim:benchmark.items.1", "path": "benchmark.items.1", "section": "benchmarks", "kind": "numeric-record", "critical": true, "text": "GPT-5.6 Sol — score: 58.9", "sourceIds": [ "openai-gpt56", "anthropic-claude-3-7-sonnet-system-card", "openai-o3-pro-system-card", "xai-grok45", "moonshot-kimi-k3", "zhipu-glm52", "alibaba-qwen37", "alibaba-qwen-2-5-coder-reasoner-paper", "swe-bench-verified-october-audit", "google-gemini35", "zhipu-glm-ocr", "baidu-ernie50", "openai-operator-sandbox-architecture", "nist-aisi-agentic-privilege-escalation-bench", "anthropic-claude-3-8-system-card", "deepseek-v3-5-architecture-report", "anthropic-claude-motion-usage-policy", "meta-muse-windows-native", "deepseek-v4", "qwen-audio3", "amd-anthropic", "reuters-aisi-agent-evals-aug5", "arcprize-opus5-jul24", "arcprize-gpt56-jul9", "artificialanalysis-v4flash-aug9", "aisi-test-time-compute", "cvpr-agentsafe-2026", "acl-safeagent-2026", "arxiv-runtime-contract-aug11", "reuters-zai-glm53-aug14", "arxiv-multi-agent-institutions-aug10", "uscc-two-loops-mar23", "reuters-unitree-us-research-aug18", "openai-third-party-cyber-aug4", "jll-emea-dc-aug3", "reuters-aisi-agent-aug20", "arc-anthropic-cot-faithfulness", "llama-3-4-moe-release", "stanford-berkeley-steering", "bird-sql-evaluation-harness", "kv-cache-mla-systems", "mit-cmu-decomposed-semantic-inversion", "swebench-pro-evaluation", "nvidia-rubin-nvl72-infrastructure", "project-glasswing-safeguards", "cmu-berkeley-agent-injection-audit", "osworld-webarena-benchmark-audit", "eth-berkeley-speculative-leak-paper", "swe-bench-verified-diagnostic-audit", "bair-stanford-quantization-safety-drift", "google-gemini-2-5-pro-reasoner-paper", "mistral-large-3-hybrid-ssm-paper", "anthropic-multimodal-prompt-smuggling", "aisi-joint-multimodal-agent-sandbox-audit", "meta-llama-4-speculative-decoding-preview", "nvidia-tensorrt-llm-fp4-blackwell", "nvidia-blackwell-ultra-nvl72-fp4", "deepseek-ascend-cann9-cluster-whitepaper" ], "citationOrigin": "explicit-claim-map", "evidenceMode": "registry-review", "evidenceNote": null, "numericValues": [ "58.9" ], "isNumeric": true, "citationCount": 58, "validSourceIds": [ "openai-gpt56", "anthropic-claude-3-7-sonnet-system-card", "openai-o3-pro-system-card", "xai-grok45", "moonshot-kimi-k3", "zhipu-glm52", "alibaba-qwen37", "alibaba-qwen-2-5-coder-reasoner-paper", "swe-bench-verified-october-audit", "google-gemini35", "zhipu-glm-ocr", "baidu-ernie50", "openai-operator-sandbox-architecture", "nist-aisi-agentic-privilege-escalation-bench", "anthropic-claude-3-8-system-card", "deepseek-v3-5-architecture-report", "anthropic-claude-motion-usage-policy", "meta-muse-windows-native", "deepseek-v4", "qwen-audio3", "amd-anthropic", "reuters-aisi-agent-evals-aug5", "arcprize-opus5-jul24", "arcprize-gpt56-jul9", "artificialanalysis-v4flash-aug9", "aisi-test-time-compute", "cvpr-agentsafe-2026", "acl-safeagent-2026", "arxiv-runtime-contract-aug11", "reuters-zai-glm53-aug14", "arxiv-multi-agent-institutions-aug10", "uscc-two-loops-mar23", "reuters-unitree-us-research-aug18", "openai-third-party-cyber-aug4", "jll-emea-dc-aug3", "reuters-aisi-agent-aug20", "arc-anthropic-cot-faithfulness", "llama-3-4-moe-release", "stanford-berkeley-steering", "bird-sql-evaluation-harness", "kv-cache-mla-systems", "mit-cmu-decomposed-semantic-inversion", "swebench-pro-evaluation", "nvidia-rubin-nvl72-infrastructure", "project-glasswing-safeguards", "cmu-berkeley-agent-injection-audit", "osworld-webarena-benchmark-audit", "eth-berkeley-speculative-leak-paper", "swe-bench-verified-diagnostic-audit", "bair-stanford-quantization-safety-drift", "google-gemini-2-5-pro-reasoner-paper", "mistral-large-3-hybrid-ssm-paper", "anthropic-multimodal-prompt-smuggling", "aisi-joint-multimodal-agent-sandbox-audit", "meta-llama-4-speculative-decoding-preview", "nvidia-tensorrt-llm-fp4-blackwell", "nvidia-blackwell-ultra-nvl72-fp4", "deepseek-ascend-cann9-cluster-whitepaper" ], "invalidSourceIds": [], "sourceDetails": [ { "id": "openai-gpt56", "title": "Introducing GPT-5.6", "publisher": "OpenAI", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true }, { "id": "anthropic-claude-3-7-sonnet-system-card", "title": "Anthropic Claude 3.7 Sonnet: Hybrid Reasoning Architecture, Extended Thinking Token Scaling, and Sub-Pixel Computer Use v2", "publisher": "Anthropic", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "openai-o3-pro-system-card", "title": "OpenAI o3-pro System Card: Frontier Reasoning, Test-Time Tree Search with Native Multi-Step Tool Execution, and Autonomous Coding Evaluations", "publisher": "OpenAI", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "xai-grok45", "title": "Grok 4.5", "publisher": "xAI", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true }, { "id": "moonshot-kimi-k3", "title": "Kimi API platform — Kimi K3 model and pricing", "publisher": "Moonshot AI", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true }, { "id": "zhipu-glm52", "title": "Zhipu BigModel — GLM-5.2 model overview", "publisher": "Zhipu AI", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true }, { "id": "alibaba-qwen37", "title": "Alibaba Cloud Model Studio — Qwen flagship models", "publisher": "Alibaba Cloud", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true }, { "id": "alibaba-qwen-2-5-coder-reasoner-paper", "title": "Qwen 2.5-Coder-32B-Reasoner: Open-Weight Test-Time Search Code Agent with Step-Level Process Reward Supervision", "publisher": "Alibaba Cloud & Qwen Team", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "swe-bench-verified-october-audit", "title": "SWE-bench Verified & Terminal-Bench October 2026 Frontier Coding Leaderboard: Test-Time Reasoning vs. Open-Weight Architectures", "publisher": "CodeEval Foundation / SWE-bench Consortium", "grade": "A", "score": 97, "freshnessStatus": "current", "critical": true }, { "id": "google-gemini35", "title": "Gemini 3.5", "publisher": "Google", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true }, { "id": "zhipu-glm-ocr", "title": "Zhipu BigModel — GLM-OCR", "publisher": "Zhipu AI", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "baidu-ernie50", "title": "Baidu Qianfan — ERNIE 5.0 model list and international price", "publisher": "Baidu AI Cloud", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true }, { "id": "openai-operator-sandbox-architecture", "title": "OpenAI Operator Architecture: MicroVM Hardware Isolation, Permission Gating, and Dual-Loop Verification for Web Agents", "publisher": "OpenAI", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "nist-aisi-agentic-privilege-escalation-bench", "title": "US Artificial Intelligence Safety Institute (NIST AISI): Standardized Test Suite for Agentic Privilege Escalation and MicroVM Containment", "publisher": "National Institute of Standards and Technology (NIST)", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "anthropic-claude-3-8-system-card", "title": "Anthropic Claude 3.8 Architecture & System Card: Computer Use v2.5, POSIX Bash Sandboxing, and SWE-bench Verified Benchmark", "publisher": "Anthropic PBC", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "deepseek-v3-5-architecture-report", "title": "DeepSeek-V3.5 Technical Report: Multi-Head Latent Attention (MLA) Scaling and DualPipe Parallelism across 2,048 Compute Nodes", "publisher": "DeepSeek AI", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "anthropic-claude-motion-usage-policy", "title": "Anthropic Claude Motion Engine, Dashboards Beta, and System Call Containment Policy", "publisher": "Anthropic PBC", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "meta-muse-windows-native", "title": "Meta Muse Desktop Architecture: Native Windows Runtime, Local Memory Buffers, and Cross-Platform Integration", "publisher": "Meta Platforms & Microsoft", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "deepseek-v4", "title": "DeepSeek V4 Preview", "publisher": "DeepSeek", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true }, { "id": "qwen-audio3", "title": "Qwen-Audio-3.0-TTS: More Multilingual, Easier to Direct", "publisher": "Alibaba Cloud / Qwen", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "amd-anthropic", "title": "AMD and Anthropic announce strategic partnership for up to 2 GW of MI450 Series GPUs", "publisher": "AMD Newsroom", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "reuters-aisi-agent-evals-aug5", "title": "OpenAI, Anthropic AI agents implicated in new security breaches", "publisher": "Reuters", "grade": "C", "score": 71, "freshnessStatus": "current", "critical": false }, { "id": "arcprize-opus5-jul24", "title": "Claude Opus 5 — ARC-AGI results", "publisher": "ARC Prize", "grade": "A", "score": 93, "freshnessStatus": "current", "critical": false }, { "id": "arcprize-gpt56-jul9", "title": "GPT-5.6 — ARC-AGI results", "publisher": "ARC Prize", "grade": "A", "score": 93, "freshnessStatus": "current", "critical": false }, { "id": "artificialanalysis-v4flash-aug9", "title": "DeepSeek V4 Flash — independent model analysis", "publisher": "Artificial Analysis", "grade": "C", "score": 71, "freshnessStatus": "current", "critical": false }, { "id": "aisi-test-time-compute", "title": "Measuring AI agents progress on multi-step cyber attack scenarios", "publisher": "UK AI Security Institute", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": false }, { "id": "cvpr-agentsafe-2026", "title": "AGENTSAFE: Benchmarking the Safety of Embodied Agents on Hazardous Instructions", "publisher": "CVPR 2026", "grade": "B", "score": 85, "freshnessStatus": "current", "critical": false }, { "id": "acl-safeagent-2026", "title": "SafeAgent: Safeguarding LLM Agents via an Automated Risk Simulator", "publisher": "ACL 2026", "grade": "B", "score": 85, "freshnessStatus": "current", "critical": false }, { "id": "arxiv-runtime-contract-aug11", "title": "Agent Safety Should Be a Runtime Contract", "publisher": "arXiv", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": false }, { "id": "reuters-zai-glm53-aug14", "title": "China's Z.ai says new model nears Anthropic's Mythos 5 in cyber-defence tests", "publisher": "Reuters", "grade": "A", "score": 93, "freshnessStatus": "current", "critical": false }, { "id": "arxiv-multi-agent-institutions-aug10", "title": "Multi-Agent AI Safety as an Institutional Design Problem", "publisher": "arXiv", "grade": "A", "score": 93, "freshnessStatus": "current", "critical": false }, { "id": "uscc-two-loops-mar23", "title": "Two Loops: How China’s Open AI Strategy Reinforces Its Industrial Dominance", "publisher": "U.S.-China Economic and Security Review Commission", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": false }, { "id": "reuters-unitree-us-research-aug18", "title": "How US military funding propelled China’s robot dogs", "publisher": "Reuters", "grade": "A", "score": 93, "freshnessStatus": "current", "critical": false }, { "id": "openai-third-party-cyber-aug4", "title": "Third-party cyber evaluations involving OpenAI models", "publisher": "OpenAI", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": false }, { "id": "jll-emea-dc-aug3", "title": "EMEA data centre mid-year 2026 report", "publisher": "JLL", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": false }, { "id": "reuters-aisi-agent-aug20", "title": "How a Texas student blew the whistle on a rogue AI hacking attempt", "publisher": "Reuters", "grade": "C", "score": 71, "freshnessStatus": "current", "critical": false }, { "id": "arc-anthropic-cot-faithfulness", "title": "Evaluating Chain-of-Thought Faithfulness Under Outcome Reinforcement Learning", "publisher": "Anthropic Research & Alignment Research Center", "grade": "A", "score": 93, "freshnessStatus": "current", "critical": false }, { "id": "llama-3-4-moe-release", "title": "Meta Llama 3.4-MoE Weights Release & Speculative Draft Heads", "publisher": "Meta AI / Hugging Face", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": false }, { "id": "stanford-berkeley-steering", "title": "Deterministic Containment of Autonomous Agent Tool Invocation via Residual Activation Steering", "publisher": "Stanford CRFM & UC Berkeley AI Research", "grade": "A", "score": 93, "freshnessStatus": "current", "critical": false }, { "id": "bird-sql-evaluation-harness", "title": "BIRD-SQL Cross-Domain Enterprise Relational Database Benchmark Leaderboard", "publisher": "BIRD-SQL Consortium / DAMO", "grade": "A", "score": 93, "freshnessStatus": "current", "critical": false }, { "id": "kv-cache-mla-systems", "title": "Multi-Head Latent Attention and 4-Bit PagedAttention Serving Implementations", "publisher": "vLLM Project & TensorRT-LLM", "grade": "A", "score": 93, "freshnessStatus": "current", "critical": false }, { "id": "mit-cmu-decomposed-semantic-inversion", "title": "Decomposed Semantic Inversion in Long-Context LLMs", "publisher": "MIT CSAIL & Carnegie Mellon University", "grade": "A", "score": 93, "freshnessStatus": "current", "critical": false }, { "id": "swebench-pro-evaluation", "title": "SWE-bench Verified Saturation Analysis and SWE-bench Pro Evaluation Methodology", "publisher": "SWE-bench Consortium", "grade": "A", "score": 93, "freshnessStatus": "current", "critical": false }, { "id": "nvidia-rubin-nvl72-infrastructure", "title": "NVIDIA Rubin NVL72 and Blackwell Ultra Rack-Scale Architecture Specifications", "publisher": "NVIDIA Corporation", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": false }, { "id": "project-glasswing-safeguards", "title": "Project Glasswing Defensive Cybersecurity Verification Framework and Enterprise Frontier Safeguards", "publisher": "Anthropic & Coalition Partners", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": false }, { "id": "cmu-berkeley-agent-injection-audit", "title": "Indirect Prompt Injection and Tool Poisoning in Autonomous Multi-Turn Agents: Empirical Vulnerability Assessment and Structural Defenses", "publisher": "Carnegie Mellon University & UC Berkeley", "grade": "A", "score": 93, "freshnessStatus": "current", "critical": false }, { "id": "osworld-webarena-benchmark-audit", "title": "OSWorld and WebArena Autonomous Agent Evaluation Leaderboards and Dynamic Environment Test Logs", "publisher": "OSWorld / WebArena Evaluation Consortium", "grade": "A", "score": 93, "freshnessStatus": "current", "critical": false }, { "id": "eth-berkeley-speculative-leak-paper", "title": "Speculative Leak: Microarchitectural Side-Channels and Cache Contention in Speculative Decoding Inference Engines (CVE-2026-38192)", "publisher": "ETH Zurich & UC Berkeley Systems and Security Lab", "grade": "A", "score": 93, "freshnessStatus": "current", "critical": false }, { "id": "swe-bench-verified-diagnostic-audit", "title": "Diagnostic Failure Modes of Autonomous Coding Agents on SWE-bench Verified: The State Synchronization Bottleneck", "publisher": "Open-Source Software Engineering Evaluation Collective", "grade": "A", "score": 93, "freshnessStatus": "current", "critical": false }, { "id": "bair-stanford-quantization-safety-drift", "title": "Asymmetric Representation Degradation: Why Extreme Quantization Erodes Safety Alignment Faster Than General Intelligence", "publisher": "Berkeley AI Research (BAIR) & Stanford Center for Research on Foundation Models (CRFM)", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": false }, { "id": "google-gemini-2-5-pro-reasoner-paper", "title": "Gemini 2.5 Pro Reasoner Technical Report: Step-Level Supervision, Monte Carlo Tree Search, and Robotics Foundation API", "publisher": "Google DeepMind", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "mistral-large-3-hybrid-ssm-paper", "title": "Mistral Large 3: Dense 145B Hybrid Selective State-Space and Transformer Model Architecture", "publisher": "Mistral AI Research", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "anthropic-multimodal-prompt-smuggling", "title": "Multi-Modal Boundary Breaches in Autonomous Execution Stacks: Exploiting Visual Representation Layers for Agent Privilege Escalation", "publisher": "Anthropic Alignment & Red-Teaming Research", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": false }, { "id": "aisi-joint-multimodal-agent-sandbox-audit", "title": "US and UK AI Safety Institutes: Joint Empirical Audit on Multimodal Visual Prompt Smuggling and Mandatory Sandbox Virtualization for Desktop Agents", "publisher": "US AI Safety Institute (NIST) / UK AI Safety Institute (DSIT)", "grade": "C", "score": 71, "freshnessStatus": "current", "critical": true }, { "id": "meta-llama-4-speculative-decoding-preview", "title": "Meta AI Engineering: Llama 4 Architectural Blueprint, Native Speculative Decoding, and MoE Routing at Scale", "publisher": "Meta AI / Fundamental AI Research", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "nvidia-tensorrt-llm-fp4-blackwell", "title": "NVIDIA TensorRT-LLM 0.16 General Availability: FP4 Microscaling Acceleration on Blackwell GB200 Systems", "publisher": "NVIDIA Technical Blog", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": false }, { "id": "nvidia-blackwell-ultra-nvl72-fp4", "title": "NVIDIA Blackwell Ultra GB200 NVL72 System Architecture: 3.2x Multi-Turn Chain-of-Thought Inference via FP4 Microscaling", "publisher": "NVIDIA Technical Blog", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "deepseek-ascend-cann9-cluster-whitepaper", "title": "DeepSeek 160,000-Accelerator Megacluster on Huawei Ascend CANN 9.0 Architecture", "publisher": "DeepSeek AI / Industry Filings", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true } ], "bestGrade": "A", "worstGrade": "C", "coverage": { "status": "supported", "reason": "At least one valid source explicitly supports this claim." } }, { "id": "claim:benchmark.items.2", "path": "benchmark.items.2", "section": "benchmarks", "kind": "numeric-record", "critical": true, "text": "Claude Opus 4.8 — score: 55.7", "sourceIds": [ "openai-gpt56", "anthropic-claude-3-7-sonnet-system-card", "openai-o3-pro-system-card", "xai-grok45", "moonshot-kimi-k3", "zhipu-glm52", "alibaba-qwen37", "alibaba-qwen-2-5-coder-reasoner-paper", "swe-bench-verified-october-audit", "google-gemini35", "zhipu-glm-ocr", "baidu-ernie50", "openai-operator-sandbox-architecture", "nist-aisi-agentic-privilege-escalation-bench", "anthropic-claude-3-8-system-card", "deepseek-v3-5-architecture-report", "anthropic-claude-motion-usage-policy", "meta-muse-windows-native", "deepseek-v4", "qwen-audio3", "amd-anthropic", "reuters-aisi-agent-evals-aug5", "arcprize-opus5-jul24", "arcprize-gpt56-jul9", "artificialanalysis-v4flash-aug9", "aisi-test-time-compute", "cvpr-agentsafe-2026", "acl-safeagent-2026", "arxiv-runtime-contract-aug11", "reuters-zai-glm53-aug14", "arxiv-multi-agent-institutions-aug10", "uscc-two-loops-mar23", "reuters-unitree-us-research-aug18", "openai-third-party-cyber-aug4", "jll-emea-dc-aug3", "reuters-aisi-agent-aug20", "arc-anthropic-cot-faithfulness", "llama-3-4-moe-release", "stanford-berkeley-steering", "bird-sql-evaluation-harness", "kv-cache-mla-systems", "mit-cmu-decomposed-semantic-inversion", "swebench-pro-evaluation", "nvidia-rubin-nvl72-infrastructure", "project-glasswing-safeguards", "cmu-berkeley-agent-injection-audit", "osworld-webarena-benchmark-audit", "eth-berkeley-speculative-leak-paper", "swe-bench-verified-diagnostic-audit", "bair-stanford-quantization-safety-drift", "google-gemini-2-5-pro-reasoner-paper", "mistral-large-3-hybrid-ssm-paper", "anthropic-multimodal-prompt-smuggling", "aisi-joint-multimodal-agent-sandbox-audit", "meta-llama-4-speculative-decoding-preview", "nvidia-tensorrt-llm-fp4-blackwell", "nvidia-blackwell-ultra-nvl72-fp4", "deepseek-ascend-cann9-cluster-whitepaper" ], "citationOrigin": "explicit-claim-map", "evidenceMode": "registry-review", "evidenceNote": null, "numericValues": [ "55.7" ], "isNumeric": true, "citationCount": 58, "validSourceIds": [ "openai-gpt56", "anthropic-claude-3-7-sonnet-system-card", "openai-o3-pro-system-card", "xai-grok45", "moonshot-kimi-k3", "zhipu-glm52", "alibaba-qwen37", "alibaba-qwen-2-5-coder-reasoner-paper", "swe-bench-verified-october-audit", "google-gemini35", "zhipu-glm-ocr", "baidu-ernie50", "openai-operator-sandbox-architecture", "nist-aisi-agentic-privilege-escalation-bench", "anthropic-claude-3-8-system-card", "deepseek-v3-5-architecture-report", "anthropic-claude-motion-usage-policy", "meta-muse-windows-native", "deepseek-v4", "qwen-audio3", "amd-anthropic", "reuters-aisi-agent-evals-aug5", "arcprize-opus5-jul24", "arcprize-gpt56-jul9", "artificialanalysis-v4flash-aug9", "aisi-test-time-compute", "cvpr-agentsafe-2026", "acl-safeagent-2026", "arxiv-runtime-contract-aug11", "reuters-zai-glm53-aug14", "arxiv-multi-agent-institutions-aug10", "uscc-two-loops-mar23", "reuters-unitree-us-research-aug18", "openai-third-party-cyber-aug4", "jll-emea-dc-aug3", "reuters-aisi-agent-aug20", "arc-anthropic-cot-faithfulness", "llama-3-4-moe-release", "stanford-berkeley-steering", "bird-sql-evaluation-harness", "kv-cache-mla-systems", "mit-cmu-decomposed-semantic-inversion", "swebench-pro-evaluation", "nvidia-rubin-nvl72-infrastructure", "project-glasswing-safeguards", "cmu-berkeley-agent-injection-audit", "osworld-webarena-benchmark-audit", "eth-berkeley-speculative-leak-paper", "swe-bench-verified-diagnostic-audit", "bair-stanford-quantization-safety-drift", "google-gemini-2-5-pro-reasoner-paper", "mistral-large-3-hybrid-ssm-paper", "anthropic-multimodal-prompt-smuggling", "aisi-joint-multimodal-agent-sandbox-audit", "meta-llama-4-speculative-decoding-preview", "nvidia-tensorrt-llm-fp4-blackwell", "nvidia-blackwell-ultra-nvl72-fp4", "deepseek-ascend-cann9-cluster-whitepaper" ], "invalidSourceIds": [], "sourceDetails": [ { "id": "openai-gpt56", "title": "Introducing GPT-5.6", "publisher": "OpenAI", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true }, { "id": "anthropic-claude-3-7-sonnet-system-card", "title": "Anthropic Claude 3.7 Sonnet: Hybrid Reasoning Architecture, Extended Thinking Token Scaling, and Sub-Pixel Computer Use v2", "publisher": "Anthropic", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "openai-o3-pro-system-card", "title": "OpenAI o3-pro System Card: Frontier Reasoning, Test-Time Tree Search with Native Multi-Step Tool Execution, and Autonomous Coding Evaluations", "publisher": "OpenAI", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "xai-grok45", "title": "Grok 4.5", "publisher": "xAI", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true }, { "id": "moonshot-kimi-k3", "title": "Kimi API platform — Kimi K3 model and pricing", "publisher": "Moonshot AI", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true }, { "id": "zhipu-glm52", "title": "Zhipu BigModel — GLM-5.2 model overview", "publisher": "Zhipu AI", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true }, { "id": "alibaba-qwen37", "title": "Alibaba Cloud Model Studio — Qwen flagship models", "publisher": "Alibaba Cloud", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true }, { "id": "alibaba-qwen-2-5-coder-reasoner-paper", "title": "Qwen 2.5-Coder-32B-Reasoner: Open-Weight Test-Time Search Code Agent with Step-Level Process Reward Supervision", "publisher": "Alibaba Cloud & Qwen Team", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "swe-bench-verified-october-audit", "title": "SWE-bench Verified & Terminal-Bench October 2026 Frontier Coding Leaderboard: Test-Time Reasoning vs. Open-Weight Architectures", "publisher": "CodeEval Foundation / SWE-bench Consortium", "grade": "A", "score": 97, "freshnessStatus": "current", "critical": true }, { "id": "google-gemini35", "title": "Gemini 3.5", "publisher": "Google", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true }, { "id": "zhipu-glm-ocr", "title": "Zhipu BigModel — GLM-OCR", "publisher": "Zhipu AI", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "baidu-ernie50", "title": "Baidu Qianfan — ERNIE 5.0 model list and international price", "publisher": "Baidu AI Cloud", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true }, { "id": "openai-operator-sandbox-architecture", "title": "OpenAI Operator Architecture: MicroVM Hardware Isolation, Permission Gating, and Dual-Loop Verification for Web Agents", "publisher": "OpenAI", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "nist-aisi-agentic-privilege-escalation-bench", "title": "US Artificial Intelligence Safety Institute (NIST AISI): Standardized Test Suite for Agentic Privilege Escalation and MicroVM Containment", "publisher": "National Institute of Standards and Technology (NIST)", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "anthropic-claude-3-8-system-card", "title": "Anthropic Claude 3.8 Architecture & System Card: Computer Use v2.5, POSIX Bash Sandboxing, and SWE-bench Verified Benchmark", "publisher": "Anthropic PBC", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "deepseek-v3-5-architecture-report", "title": "DeepSeek-V3.5 Technical Report: Multi-Head Latent Attention (MLA) Scaling and DualPipe Parallelism across 2,048 Compute Nodes", "publisher": "DeepSeek AI", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "anthropic-claude-motion-usage-policy", "title": "Anthropic Claude Motion Engine, Dashboards Beta, and System Call Containment Policy", "publisher": "Anthropic PBC", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "meta-muse-windows-native", "title": "Meta Muse Desktop Architecture: Native Windows Runtime, Local Memory Buffers, and Cross-Platform Integration", "publisher": "Meta Platforms & Microsoft", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "deepseek-v4", "title": "DeepSeek V4 Preview", "publisher": "DeepSeek", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true }, { "id": "qwen-audio3", "title": "Qwen-Audio-3.0-TTS: More Multilingual, Easier to Direct", "publisher": "Alibaba Cloud / Qwen", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "amd-anthropic", "title": "AMD and Anthropic announce strategic partnership for up to 2 GW of MI450 Series GPUs", "publisher": "AMD Newsroom", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "reuters-aisi-agent-evals-aug5", "title": "OpenAI, Anthropic AI agents implicated in new security breaches", "publisher": "Reuters", "grade": "C", "score": 71, "freshnessStatus": "current", "critical": false }, { "id": "arcprize-opus5-jul24", "title": "Claude Opus 5 — ARC-AGI results", "publisher": "ARC Prize", "grade": "A", "score": 93, "freshnessStatus": "current", "critical": false }, { "id": "arcprize-gpt56-jul9", "title": "GPT-5.6 — ARC-AGI results", "publisher": "ARC Prize", "grade": "A", "score": 93, "freshnessStatus": "current", "critical": false }, { "id": "artificialanalysis-v4flash-aug9", "title": "DeepSeek V4 Flash — independent model analysis", "publisher": "Artificial Analysis", "grade": "C", "score": 71, "freshnessStatus": "current", "critical": false }, { "id": "aisi-test-time-compute", "title": "Measuring AI agents progress on multi-step cyber attack scenarios", "publisher": "UK AI Security Institute", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": false }, { "id": "cvpr-agentsafe-2026", "title": "AGENTSAFE: Benchmarking the Safety of Embodied Agents on Hazardous Instructions", "publisher": "CVPR 2026", "grade": "B", "score": 85, "freshnessStatus": "current", "critical": false }, { "id": "acl-safeagent-2026", "title": "SafeAgent: Safeguarding LLM Agents via an Automated Risk Simulator", "publisher": "ACL 2026", "grade": "B", "score": 85, "freshnessStatus": "current", "critical": false }, { "id": "arxiv-runtime-contract-aug11", "title": "Agent Safety Should Be a Runtime Contract", "publisher": "arXiv", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": false }, { "id": "reuters-zai-glm53-aug14", "title": "China's Z.ai says new model nears Anthropic's Mythos 5 in cyber-defence tests", "publisher": "Reuters", "grade": "A", "score": 93, "freshnessStatus": "current", "critical": false }, { "id": "arxiv-multi-agent-institutions-aug10", "title": "Multi-Agent AI Safety as an Institutional Design Problem", "publisher": "arXiv", "grade": "A", "score": 93, "freshnessStatus": "current", "critical": false }, { "id": "uscc-two-loops-mar23", "title": "Two Loops: How China’s Open AI Strategy Reinforces Its Industrial Dominance", "publisher": "U.S.-China Economic and Security Review Commission", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": false }, { "id": "reuters-unitree-us-research-aug18", "title": "How US military funding propelled China’s robot dogs", "publisher": "Reuters", "grade": "A", "score": 93, "freshnessStatus": "current", "critical": false }, { "id": "openai-third-party-cyber-aug4", "title": "Third-party cyber evaluations involving OpenAI models", "publisher": "OpenAI", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": false }, { "id": "jll-emea-dc-aug3", "title": "EMEA data centre mid-year 2026 report", "publisher": "JLL", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": false }, { "id": "reuters-aisi-agent-aug20", "title": "How a Texas student blew the whistle on a rogue AI hacking attempt", "publisher": "Reuters", "grade": "C", "score": 71, "freshnessStatus": "current", "critical": false }, { "id": "arc-anthropic-cot-faithfulness", "title": "Evaluating Chain-of-Thought Faithfulness Under Outcome Reinforcement Learning", "publisher": "Anthropic Research & Alignment Research Center", "grade": "A", "score": 93, "freshnessStatus": "current", "critical": false }, { "id": "llama-3-4-moe-release", "title": "Meta Llama 3.4-MoE Weights Release & Speculative Draft Heads", "publisher": "Meta AI / Hugging Face", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": false }, { "id": "stanford-berkeley-steering", "title": "Deterministic Containment of Autonomous Agent Tool Invocation via Residual Activation Steering", "publisher": "Stanford CRFM & UC Berkeley AI Research", "grade": "A", "score": 93, "freshnessStatus": "current", "critical": false }, { "id": "bird-sql-evaluation-harness", "title": "BIRD-SQL Cross-Domain Enterprise Relational Database Benchmark Leaderboard", "publisher": "BIRD-SQL Consortium / DAMO", "grade": "A", "score": 93, "freshnessStatus": "current", "critical": false }, { "id": "kv-cache-mla-systems", "title": "Multi-Head Latent Attention and 4-Bit PagedAttention Serving Implementations", "publisher": "vLLM Project & TensorRT-LLM", "grade": "A", "score": 93, "freshnessStatus": "current", "critical": false }, { "id": "mit-cmu-decomposed-semantic-inversion", "title": "Decomposed Semantic Inversion in Long-Context LLMs", "publisher": "MIT CSAIL & Carnegie Mellon University", "grade": "A", "score": 93, "freshnessStatus": "current", "critical": false }, { "id": "swebench-pro-evaluation", "title": "SWE-bench Verified Saturation Analysis and SWE-bench Pro Evaluation Methodology", "publisher": "SWE-bench Consortium", "grade": "A", "score": 93, "freshnessStatus": "current", "critical": false }, { "id": "nvidia-rubin-nvl72-infrastructure", "title": "NVIDIA Rubin NVL72 and Blackwell Ultra Rack-Scale Architecture Specifications", "publisher": "NVIDIA Corporation", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": false }, { "id": "project-glasswing-safeguards", "title": "Project Glasswing Defensive Cybersecurity Verification Framework and Enterprise Frontier Safeguards", "publisher": "Anthropic & Coalition Partners", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": false }, { "id": "cmu-berkeley-agent-injection-audit", "title": "Indirect Prompt Injection and Tool Poisoning in Autonomous Multi-Turn Agents: Empirical Vulnerability Assessment and Structural Defenses", "publisher": "Carnegie Mellon University & UC Berkeley", "grade": "A", "score": 93, "freshnessStatus": "current", "critical": false }, { "id": "osworld-webarena-benchmark-audit", "title": "OSWorld and WebArena Autonomous Agent Evaluation Leaderboards and Dynamic Environment Test Logs", "publisher": "OSWorld / WebArena Evaluation Consortium", "grade": "A", "score": 93, "freshnessStatus": "current", "critical": false }, { "id": "eth-berkeley-speculative-leak-paper", "title": "Speculative Leak: Microarchitectural Side-Channels and Cache Contention in Speculative Decoding Inference Engines (CVE-2026-38192)", "publisher": "ETH Zurich & UC Berkeley Systems and Security Lab", "grade": "A", "score": 93, "freshnessStatus": "current", "critical": false }, { "id": "swe-bench-verified-diagnostic-audit", "title": "Diagnostic Failure Modes of Autonomous Coding Agents on SWE-bench Verified: The State Synchronization Bottleneck", "publisher": "Open-Source Software Engineering Evaluation Collective", "grade": "A", "score": 93, "freshnessStatus": "current", "critical": false }, { "id": "bair-stanford-quantization-safety-drift", "title": "Asymmetric Representation Degradation: Why Extreme Quantization Erodes Safety Alignment Faster Than General Intelligence", "publisher": "Berkeley AI Research (BAIR) & Stanford Center for Research on Foundation Models (CRFM)", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": false }, { "id": "google-gemini-2-5-pro-reasoner-paper", "title": "Gemini 2.5 Pro Reasoner Technical Report: Step-Level Supervision, Monte Carlo Tree Search, and Robotics Foundation API", "publisher": "Google DeepMind", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "mistral-large-3-hybrid-ssm-paper", "title": "Mistral Large 3: Dense 145B Hybrid Selective State-Space and Transformer Model Architecture", "publisher": "Mistral AI Research", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "anthropic-multimodal-prompt-smuggling", "title": "Multi-Modal Boundary Breaches in Autonomous Execution Stacks: Exploiting Visual Representation Layers for Agent Privilege Escalation", "publisher": "Anthropic Alignment & Red-Teaming Research", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": false }, { "id": "aisi-joint-multimodal-agent-sandbox-audit", "title": "US and UK AI Safety Institutes: Joint Empirical Audit on Multimodal Visual Prompt Smuggling and Mandatory Sandbox Virtualization for Desktop Agents", "publisher": "US AI Safety Institute (NIST) / UK AI Safety Institute (DSIT)", "grade": "C", "score": 71, "freshnessStatus": "current", "critical": true }, { "id": "meta-llama-4-speculative-decoding-preview", "title": "Meta AI Engineering: Llama 4 Architectural Blueprint, Native Speculative Decoding, and MoE Routing at Scale", "publisher": "Meta AI / Fundamental AI Research", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "nvidia-tensorrt-llm-fp4-blackwell", "title": "NVIDIA TensorRT-LLM 0.16 General Availability: FP4 Microscaling Acceleration on Blackwell GB200 Systems", "publisher": "NVIDIA Technical Blog", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": false }, { "id": "nvidia-blackwell-ultra-nvl72-fp4", "title": "NVIDIA Blackwell Ultra GB200 NVL72 System Architecture: 3.2x Multi-Turn Chain-of-Thought Inference via FP4 Microscaling", "publisher": "NVIDIA Technical Blog", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "deepseek-ascend-cann9-cluster-whitepaper", "title": "DeepSeek 160,000-Accelerator Megacluster on Huawei Ascend CANN 9.0 Architecture", "publisher": "DeepSeek AI / Industry Filings", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true } ], "bestGrade": "A", "worstGrade": "C", "coverage": { "status": "supported", "reason": "At least one valid source explicitly supports this claim." } }, { "id": "claim:benchmark.items.3", "path": "benchmark.items.3", "section": "benchmarks", "kind": "numeric-record", "critical": true, "text": "GPT-5.5 — score: 54.8", "sourceIds": [ "openai-gpt56", "anthropic-claude-3-7-sonnet-system-card", "openai-o3-pro-system-card", "xai-grok45", "moonshot-kimi-k3", "zhipu-glm52", "alibaba-qwen37", "alibaba-qwen-2-5-coder-reasoner-paper", "swe-bench-verified-october-audit", "google-gemini35", "zhipu-glm-ocr", "baidu-ernie50", "openai-operator-sandbox-architecture", "nist-aisi-agentic-privilege-escalation-bench", "anthropic-claude-3-8-system-card", "deepseek-v3-5-architecture-report", "anthropic-claude-motion-usage-policy", "meta-muse-windows-native", "deepseek-v4", "qwen-audio3", "amd-anthropic", "reuters-aisi-agent-evals-aug5", "arcprize-opus5-jul24", "arcprize-gpt56-jul9", "artificialanalysis-v4flash-aug9", "aisi-test-time-compute", "cvpr-agentsafe-2026", "acl-safeagent-2026", "arxiv-runtime-contract-aug11", "reuters-zai-glm53-aug14", "arxiv-multi-agent-institutions-aug10", "uscc-two-loops-mar23", "reuters-unitree-us-research-aug18", "openai-third-party-cyber-aug4", "jll-emea-dc-aug3", "reuters-aisi-agent-aug20", "arc-anthropic-cot-faithfulness", "llama-3-4-moe-release", "stanford-berkeley-steering", "bird-sql-evaluation-harness", "kv-cache-mla-systems", "mit-cmu-decomposed-semantic-inversion", "swebench-pro-evaluation", "nvidia-rubin-nvl72-infrastructure", "project-glasswing-safeguards", "cmu-berkeley-agent-injection-audit", "osworld-webarena-benchmark-audit", "eth-berkeley-speculative-leak-paper", "swe-bench-verified-diagnostic-audit", "bair-stanford-quantization-safety-drift", "google-gemini-2-5-pro-reasoner-paper", "mistral-large-3-hybrid-ssm-paper", "anthropic-multimodal-prompt-smuggling", "aisi-joint-multimodal-agent-sandbox-audit", "meta-llama-4-speculative-decoding-preview", "nvidia-tensorrt-llm-fp4-blackwell", "nvidia-blackwell-ultra-nvl72-fp4", "deepseek-ascend-cann9-cluster-whitepaper" ], "citationOrigin": "explicit-claim-map", "evidenceMode": "registry-review", "evidenceNote": null, "numericValues": [ "54.8" ], "isNumeric": true, "citationCount": 58, "validSourceIds": [ "openai-gpt56", "anthropic-claude-3-7-sonnet-system-card", "openai-o3-pro-system-card", "xai-grok45", "moonshot-kimi-k3", "zhipu-glm52", "alibaba-qwen37", "alibaba-qwen-2-5-coder-reasoner-paper", "swe-bench-verified-october-audit", "google-gemini35", "zhipu-glm-ocr", "baidu-ernie50", "openai-operator-sandbox-architecture", "nist-aisi-agentic-privilege-escalation-bench", "anthropic-claude-3-8-system-card", "deepseek-v3-5-architecture-report", "anthropic-claude-motion-usage-policy", "meta-muse-windows-native", "deepseek-v4", "qwen-audio3", "amd-anthropic", "reuters-aisi-agent-evals-aug5", "arcprize-opus5-jul24", "arcprize-gpt56-jul9", "artificialanalysis-v4flash-aug9", "aisi-test-time-compute", "cvpr-agentsafe-2026", "acl-safeagent-2026", "arxiv-runtime-contract-aug11", "reuters-zai-glm53-aug14", "arxiv-multi-agent-institutions-aug10", "uscc-two-loops-mar23", "reuters-unitree-us-research-aug18", "openai-third-party-cyber-aug4", "jll-emea-dc-aug3", "reuters-aisi-agent-aug20", "arc-anthropic-cot-faithfulness", "llama-3-4-moe-release", "stanford-berkeley-steering", "bird-sql-evaluation-harness", "kv-cache-mla-systems", "mit-cmu-decomposed-semantic-inversion", "swebench-pro-evaluation", "nvidia-rubin-nvl72-infrastructure", "project-glasswing-safeguards", "cmu-berkeley-agent-injection-audit", "osworld-webarena-benchmark-audit", "eth-berkeley-speculative-leak-paper", "swe-bench-verified-diagnostic-audit", "bair-stanford-quantization-safety-drift", "google-gemini-2-5-pro-reasoner-paper", "mistral-large-3-hybrid-ssm-paper", "anthropic-multimodal-prompt-smuggling", "aisi-joint-multimodal-agent-sandbox-audit", "meta-llama-4-speculative-decoding-preview", "nvidia-tensorrt-llm-fp4-blackwell", "nvidia-blackwell-ultra-nvl72-fp4", "deepseek-ascend-cann9-cluster-whitepaper" ], "invalidSourceIds": [], "sourceDetails": [ { "id": "openai-gpt56", "title": "Introducing GPT-5.6", "publisher": "OpenAI", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true }, { "id": "anthropic-claude-3-7-sonnet-system-card", "title": "Anthropic Claude 3.7 Sonnet: Hybrid Reasoning Architecture, Extended Thinking Token Scaling, and Sub-Pixel Computer Use v2", "publisher": "Anthropic", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "openai-o3-pro-system-card", "title": "OpenAI o3-pro System Card: Frontier Reasoning, Test-Time Tree Search with Native Multi-Step Tool Execution, and Autonomous Coding Evaluations", "publisher": "OpenAI", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "xai-grok45", "title": "Grok 4.5", "publisher": "xAI", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true }, { "id": "moonshot-kimi-k3", "title": "Kimi API platform — Kimi K3 model and pricing", "publisher": "Moonshot AI", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true }, { "id": "zhipu-glm52", "title": "Zhipu BigModel — GLM-5.2 model overview", "publisher": "Zhipu AI", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true }, { "id": "alibaba-qwen37", "title": "Alibaba Cloud Model Studio — Qwen flagship models", "publisher": "Alibaba Cloud", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true }, { "id": "alibaba-qwen-2-5-coder-reasoner-paper", "title": "Qwen 2.5-Coder-32B-Reasoner: Open-Weight Test-Time Search Code Agent with Step-Level Process Reward Supervision", "publisher": "Alibaba Cloud & Qwen Team", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "swe-bench-verified-october-audit", "title": "SWE-bench Verified & Terminal-Bench October 2026 Frontier Coding Leaderboard: Test-Time Reasoning vs. Open-Weight Architectures", "publisher": "CodeEval Foundation / SWE-bench Consortium", "grade": "A", "score": 97, "freshnessStatus": "current", "critical": true }, { "id": "google-gemini35", "title": "Gemini 3.5", "publisher": "Google", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true }, { "id": "zhipu-glm-ocr", "title": "Zhipu BigModel — GLM-OCR", "publisher": "Zhipu AI", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "baidu-ernie50", "title": "Baidu Qianfan — ERNIE 5.0 model list and international price", "publisher": "Baidu AI Cloud", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true }, { "id": "openai-operator-sandbox-architecture", "title": "OpenAI Operator Architecture: MicroVM Hardware Isolation, Permission Gating, and Dual-Loop Verification for Web Agents", "publisher": "OpenAI", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "nist-aisi-agentic-privilege-escalation-bench", "title": "US Artificial Intelligence Safety Institute (NIST AISI): Standardized Test Suite for Agentic Privilege Escalation and MicroVM Containment", "publisher": "National Institute of Standards and Technology (NIST)", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "anthropic-claude-3-8-system-card", "title": "Anthropic Claude 3.8 Architecture & System Card: Computer Use v2.5, POSIX Bash Sandboxing, and SWE-bench Verified Benchmark", "publisher": "Anthropic PBC", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "deepseek-v3-5-architecture-report", "title": "DeepSeek-V3.5 Technical Report: Multi-Head Latent Attention (MLA) Scaling and DualPipe Parallelism across 2,048 Compute Nodes", "publisher": "DeepSeek AI", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "anthropic-claude-motion-usage-policy", "title": "Anthropic Claude Motion Engine, Dashboards Beta, and System Call Containment Policy", "publisher": "Anthropic PBC", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "meta-muse-windows-native", "title": "Meta Muse Desktop Architecture: Native Windows Runtime, Local Memory Buffers, and Cross-Platform Integration", "publisher": "Meta Platforms & Microsoft", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "deepseek-v4", "title": "DeepSeek V4 Preview", "publisher": "DeepSeek", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true }, { "id": "qwen-audio3", "title": "Qwen-Audio-3.0-TTS: More Multilingual, Easier to Direct", "publisher": "Alibaba Cloud / Qwen", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "amd-anthropic", "title": "AMD and Anthropic announce strategic partnership for up to 2 GW of MI450 Series GPUs", "publisher": "AMD Newsroom", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "reuters-aisi-agent-evals-aug5", "title": "OpenAI, Anthropic AI agents implicated in new security breaches", "publisher": "Reuters", "grade": "C", "score": 71, "freshnessStatus": "current", "critical": false }, { "id": "arcprize-opus5-jul24", "title": "Claude Opus 5 — ARC-AGI results", "publisher": "ARC Prize", "grade": "A", "score": 93, "freshnessStatus": "current", "critical": false }, { "id": "arcprize-gpt56-jul9", "title": "GPT-5.6 — ARC-AGI results", "publisher": "ARC Prize", "grade": "A", "score": 93, "freshnessStatus": "current", "critical": false }, { "id": "artificialanalysis-v4flash-aug9", "title": "DeepSeek V4 Flash — independent model analysis", "publisher": "Artificial Analysis", "grade": "C", "score": 71, "freshnessStatus": "current", "critical": false }, { "id": "aisi-test-time-compute", "title": "Measuring AI agents progress on multi-step cyber attack scenarios", "publisher": "UK AI Security Institute", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": false }, { "id": "cvpr-agentsafe-2026", "title": "AGENTSAFE: Benchmarking the Safety of Embodied Agents on Hazardous Instructions", "publisher": "CVPR 2026", "grade": "B", "score": 85, "freshnessStatus": "current", "critical": false }, { "id": "acl-safeagent-2026", "title": "SafeAgent: Safeguarding LLM Agents via an Automated Risk Simulator", "publisher": "ACL 2026", "grade": "B", "score": 85, "freshnessStatus": "current", "critical": false }, { "id": "arxiv-runtime-contract-aug11", "title": "Agent Safety Should Be a Runtime Contract", "publisher": "arXiv", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": false }, { "id": "reuters-zai-glm53-aug14", "title": "China's Z.ai says new model nears Anthropic's Mythos 5 in cyber-defence tests", "publisher": "Reuters", "grade": "A", "score": 93, "freshnessStatus": "current", "critical": false }, { "id": "arxiv-multi-agent-institutions-aug10", "title": "Multi-Agent AI Safety as an Institutional Design Problem", "publisher": "arXiv", "grade": "A", "score": 93, "freshnessStatus": "current", "critical": false }, { "id": "uscc-two-loops-mar23", "title": "Two Loops: How China’s Open AI Strategy Reinforces Its Industrial Dominance", "publisher": "U.S.-China Economic and Security Review Commission", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": false }, { "id": "reuters-unitree-us-research-aug18", "title": "How US military funding propelled China’s robot dogs", "publisher": "Reuters", "grade": "A", "score": 93, "freshnessStatus": "current", "critical": false }, { "id": "openai-third-party-cyber-aug4", "title": "Third-party cyber evaluations involving OpenAI models", "publisher": "OpenAI", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": false }, { "id": "jll-emea-dc-aug3", "title": "EMEA data centre mid-year 2026 report", "publisher": "JLL", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": false }, { "id": "reuters-aisi-agent-aug20", "title": "How a Texas student blew the whistle on a rogue AI hacking attempt", "publisher": "Reuters", "grade": "C", "score": 71, "freshnessStatus": "current", "critical": false }, { "id": "arc-anthropic-cot-faithfulness", "title": "Evaluating Chain-of-Thought Faithfulness Under Outcome Reinforcement Learning", "publisher": "Anthropic Research & Alignment Research Center", "grade": "A", "score": 93, "freshnessStatus": "current", "critical": false }, { "id": "llama-3-4-moe-release", "title": "Meta Llama 3.4-MoE Weights Release & Speculative Draft Heads", "publisher": "Meta AI / Hugging Face", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": false }, { "id": "stanford-berkeley-steering", "title": "Deterministic Containment of Autonomous Agent Tool Invocation via Residual Activation Steering", "publisher": "Stanford CRFM & UC Berkeley AI Research", "grade": "A", "score": 93, "freshnessStatus": "current", "critical": false }, { "id": "bird-sql-evaluation-harness", "title": "BIRD-SQL Cross-Domain Enterprise Relational Database Benchmark Leaderboard", "publisher": "BIRD-SQL Consortium / DAMO", "grade": "A", "score": 93, "freshnessStatus": "current", "critical": false }, { "id": "kv-cache-mla-systems", "title": "Multi-Head Latent Attention and 4-Bit PagedAttention Serving Implementations", "publisher": "vLLM Project & TensorRT-LLM", "grade": "A", "score": 93, "freshnessStatus": "current", "critical": false }, { "id": "mit-cmu-decomposed-semantic-inversion", "title": "Decomposed Semantic Inversion in Long-Context LLMs", "publisher": "MIT CSAIL & Carnegie Mellon University", "grade": "A", "score": 93, "freshnessStatus": "current", "critical": false }, { "id": "swebench-pro-evaluation", "title": "SWE-bench Verified Saturation Analysis and SWE-bench Pro Evaluation Methodology", "publisher": "SWE-bench Consortium", "grade": "A", "score": 93, "freshnessStatus": "current", "critical": false }, { "id": "nvidia-rubin-nvl72-infrastructure", "title": "NVIDIA Rubin NVL72 and Blackwell Ultra Rack-Scale Architecture Specifications", "publisher": "NVIDIA Corporation", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": false }, { "id": "project-glasswing-safeguards", "title": "Project Glasswing Defensive Cybersecurity Verification Framework and Enterprise Frontier Safeguards", "publisher": "Anthropic & Coalition Partners", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": false }, { "id": "cmu-berkeley-agent-injection-audit", "title": "Indirect Prompt Injection and Tool Poisoning in Autonomous Multi-Turn Agents: Empirical Vulnerability Assessment and Structural Defenses", "publisher": "Carnegie Mellon University & UC Berkeley", "grade": "A", "score": 93, "freshnessStatus": "current", "critical": false }, { "id": "osworld-webarena-benchmark-audit", "title": "OSWorld and WebArena Autonomous Agent Evaluation Leaderboards and Dynamic Environment Test Logs", "publisher": "OSWorld / WebArena Evaluation Consortium", "grade": "A", "score": 93, "freshnessStatus": "current", "critical": false }, { "id": "eth-berkeley-speculative-leak-paper", "title": "Speculative Leak: Microarchitectural Side-Channels and Cache Contention in Speculative Decoding Inference Engines (CVE-2026-38192)", "publisher": "ETH Zurich & UC Berkeley Systems and Security Lab", "grade": "A", "score": 93, "freshnessStatus": "current", "critical": false }, { "id": "swe-bench-verified-diagnostic-audit", "title": "Diagnostic Failure Modes of Autonomous Coding Agents on SWE-bench Verified: The State Synchronization Bottleneck", "publisher": "Open-Source Software Engineering Evaluation Collective", "grade": "A", "score": 93, "freshnessStatus": "current", "critical": false }, { "id": "bair-stanford-quantization-safety-drift", "title": "Asymmetric Representation Degradation: Why Extreme Quantization Erodes Safety Alignment Faster Than General Intelligence", "publisher": "Berkeley AI Research (BAIR) & Stanford Center for Research on Foundation Models (CRFM)", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": false }, { "id": "google-gemini-2-5-pro-reasoner-paper", "title": "Gemini 2.5 Pro Reasoner Technical Report: Step-Level Supervision, Monte Carlo Tree Search, and Robotics Foundation API", "publisher": "Google DeepMind", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "mistral-large-3-hybrid-ssm-paper", "title": "Mistral Large 3: Dense 145B Hybrid Selective State-Space and Transformer Model Architecture", "publisher": "Mistral AI Research", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "anthropic-multimodal-prompt-smuggling", "title": "Multi-Modal Boundary Breaches in Autonomous Execution Stacks: Exploiting Visual Representation Layers for Agent Privilege Escalation", "publisher": "Anthropic Alignment & Red-Teaming Research", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": false }, { "id": "aisi-joint-multimodal-agent-sandbox-audit", "title": "US and UK AI Safety Institutes: Joint Empirical Audit on Multimodal Visual Prompt Smuggling and Mandatory Sandbox Virtualization for Desktop Agents", "publisher": "US AI Safety Institute (NIST) / UK AI Safety Institute (DSIT)", "grade": "C", "score": 71, "freshnessStatus": "current", "critical": true }, { "id": "meta-llama-4-speculative-decoding-preview", "title": "Meta AI Engineering: Llama 4 Architectural Blueprint, Native Speculative Decoding, and MoE Routing at Scale", "publisher": "Meta AI / Fundamental AI Research", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "nvidia-tensorrt-llm-fp4-blackwell", "title": "NVIDIA TensorRT-LLM 0.16 General Availability: FP4 Microscaling Acceleration on Blackwell GB200 Systems", "publisher": "NVIDIA Technical Blog", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": false }, { "id": "nvidia-blackwell-ultra-nvl72-fp4", "title": "NVIDIA Blackwell Ultra GB200 NVL72 System Architecture: 3.2x Multi-Turn Chain-of-Thought Inference via FP4 Microscaling", "publisher": "NVIDIA Technical Blog", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "deepseek-ascend-cann9-cluster-whitepaper", "title": "DeepSeek 160,000-Accelerator Megacluster on Huawei Ascend CANN 9.0 Architecture", "publisher": "DeepSeek AI / Industry Filings", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true } ], "bestGrade": "A", "worstGrade": "C", "coverage": { "status": "supported", "reason": "At least one valid source explicitly supports this claim." } }, { "id": "claim:benchmark.items.4", "path": "benchmark.items.4", "section": "benchmarks", "kind": "numeric-record", "critical": true, "text": "GPT-5.6 Terra — score: 55", "sourceIds": [ "openai-gpt56", "anthropic-claude-3-7-sonnet-system-card", "openai-o3-pro-system-card", "xai-grok45", "moonshot-kimi-k3", "zhipu-glm52", "alibaba-qwen37", "alibaba-qwen-2-5-coder-reasoner-paper", "swe-bench-verified-october-audit", "google-gemini35", "zhipu-glm-ocr", "baidu-ernie50", "openai-operator-sandbox-architecture", "nist-aisi-agentic-privilege-escalation-bench", "anthropic-claude-3-8-system-card", "deepseek-v3-5-architecture-report", "anthropic-claude-motion-usage-policy", "meta-muse-windows-native", "deepseek-v4", "qwen-audio3", "amd-anthropic", "reuters-aisi-agent-evals-aug5", "arcprize-opus5-jul24", "arcprize-gpt56-jul9", "artificialanalysis-v4flash-aug9", "aisi-test-time-compute", "cvpr-agentsafe-2026", "acl-safeagent-2026", "arxiv-runtime-contract-aug11", "reuters-zai-glm53-aug14", "arxiv-multi-agent-institutions-aug10", "uscc-two-loops-mar23", "reuters-unitree-us-research-aug18", "openai-third-party-cyber-aug4", "jll-emea-dc-aug3", "reuters-aisi-agent-aug20", "arc-anthropic-cot-faithfulness", "llama-3-4-moe-release", "stanford-berkeley-steering", "bird-sql-evaluation-harness", "kv-cache-mla-systems", "mit-cmu-decomposed-semantic-inversion", "swebench-pro-evaluation", "nvidia-rubin-nvl72-infrastructure", "project-glasswing-safeguards", "cmu-berkeley-agent-injection-audit", "osworld-webarena-benchmark-audit", "eth-berkeley-speculative-leak-paper", "swe-bench-verified-diagnostic-audit", "bair-stanford-quantization-safety-drift", "google-gemini-2-5-pro-reasoner-paper", "mistral-large-3-hybrid-ssm-paper", "anthropic-multimodal-prompt-smuggling", "aisi-joint-multimodal-agent-sandbox-audit", "meta-llama-4-speculative-decoding-preview", "nvidia-tensorrt-llm-fp4-blackwell", "nvidia-blackwell-ultra-nvl72-fp4", "deepseek-ascend-cann9-cluster-whitepaper" ], "citationOrigin": "explicit-claim-map", "evidenceMode": "registry-review", "evidenceNote": null, "numericValues": [ "55" ], "isNumeric": true, "citationCount": 58, "validSourceIds": [ "openai-gpt56", "anthropic-claude-3-7-sonnet-system-card", "openai-o3-pro-system-card", "xai-grok45", "moonshot-kimi-k3", "zhipu-glm52", "alibaba-qwen37", "alibaba-qwen-2-5-coder-reasoner-paper", "swe-bench-verified-october-audit", "google-gemini35", "zhipu-glm-ocr", "baidu-ernie50", "openai-operator-sandbox-architecture", "nist-aisi-agentic-privilege-escalation-bench", "anthropic-claude-3-8-system-card", "deepseek-v3-5-architecture-report", "anthropic-claude-motion-usage-policy", "meta-muse-windows-native", "deepseek-v4", "qwen-audio3", "amd-anthropic", "reuters-aisi-agent-evals-aug5", "arcprize-opus5-jul24", "arcprize-gpt56-jul9", "artificialanalysis-v4flash-aug9", "aisi-test-time-compute", "cvpr-agentsafe-2026", "acl-safeagent-2026", "arxiv-runtime-contract-aug11", "reuters-zai-glm53-aug14", "arxiv-multi-agent-institutions-aug10", "uscc-two-loops-mar23", "reuters-unitree-us-research-aug18", "openai-third-party-cyber-aug4", "jll-emea-dc-aug3", "reuters-aisi-agent-aug20", "arc-anthropic-cot-faithfulness", "llama-3-4-moe-release", "stanford-berkeley-steering", "bird-sql-evaluation-harness", "kv-cache-mla-systems", "mit-cmu-decomposed-semantic-inversion", "swebench-pro-evaluation", "nvidia-rubin-nvl72-infrastructure", "project-glasswing-safeguards", "cmu-berkeley-agent-injection-audit", "osworld-webarena-benchmark-audit", "eth-berkeley-speculative-leak-paper", "swe-bench-verified-diagnostic-audit", "bair-stanford-quantization-safety-drift", "google-gemini-2-5-pro-reasoner-paper", "mistral-large-3-hybrid-ssm-paper", "anthropic-multimodal-prompt-smuggling", "aisi-joint-multimodal-agent-sandbox-audit", "meta-llama-4-speculative-decoding-preview", "nvidia-tensorrt-llm-fp4-blackwell", "nvidia-blackwell-ultra-nvl72-fp4", "deepseek-ascend-cann9-cluster-whitepaper" ], "invalidSourceIds": [], "sourceDetails": [ { "id": "openai-gpt56", "title": "Introducing GPT-5.6", "publisher": "OpenAI", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true }, { "id": "anthropic-claude-3-7-sonnet-system-card", "title": "Anthropic Claude 3.7 Sonnet: Hybrid Reasoning Architecture, Extended Thinking Token Scaling, and Sub-Pixel Computer Use v2", "publisher": "Anthropic", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "openai-o3-pro-system-card", "title": "OpenAI o3-pro System Card: Frontier Reasoning, Test-Time Tree Search with Native Multi-Step Tool Execution, and Autonomous Coding Evaluations", "publisher": "OpenAI", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "xai-grok45", "title": "Grok 4.5", "publisher": "xAI", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true }, { "id": "moonshot-kimi-k3", "title": "Kimi API platform — Kimi K3 model and pricing", "publisher": "Moonshot AI", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true }, { "id": "zhipu-glm52", "title": "Zhipu BigModel — GLM-5.2 model overview", "publisher": "Zhipu AI", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true }, { "id": "alibaba-qwen37", "title": "Alibaba Cloud Model Studio — Qwen flagship models", "publisher": "Alibaba Cloud", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true }, { "id": "alibaba-qwen-2-5-coder-reasoner-paper", "title": "Qwen 2.5-Coder-32B-Reasoner: Open-Weight Test-Time Search Code Agent with Step-Level Process Reward Supervision", "publisher": "Alibaba Cloud & Qwen Team", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "swe-bench-verified-october-audit", "title": "SWE-bench Verified & Terminal-Bench October 2026 Frontier Coding Leaderboard: Test-Time Reasoning vs. Open-Weight Architectures", "publisher": "CodeEval Foundation / SWE-bench Consortium", "grade": "A", "score": 97, "freshnessStatus": "current", "critical": true }, { "id": "google-gemini35", "title": "Gemini 3.5", "publisher": "Google", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true }, { "id": "zhipu-glm-ocr", "title": "Zhipu BigModel — GLM-OCR", "publisher": "Zhipu AI", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "baidu-ernie50", "title": "Baidu Qianfan — ERNIE 5.0 model list and international price", "publisher": "Baidu AI Cloud", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true }, { "id": "openai-operator-sandbox-architecture", "title": "OpenAI Operator Architecture: MicroVM Hardware Isolation, Permission Gating, and Dual-Loop Verification for Web Agents", "publisher": "OpenAI", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "nist-aisi-agentic-privilege-escalation-bench", "title": "US Artificial Intelligence Safety Institute (NIST AISI): Standardized Test Suite for Agentic Privilege Escalation and MicroVM Containment", "publisher": "National Institute of Standards and Technology (NIST)", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "anthropic-claude-3-8-system-card", "title": "Anthropic Claude 3.8 Architecture & System Card: Computer Use v2.5, POSIX Bash Sandboxing, and SWE-bench Verified Benchmark", "publisher": "Anthropic PBC", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "deepseek-v3-5-architecture-report", "title": "DeepSeek-V3.5 Technical Report: Multi-Head Latent Attention (MLA) Scaling and DualPipe Parallelism across 2,048 Compute Nodes", "publisher": "DeepSeek AI", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "anthropic-claude-motion-usage-policy", "title": "Anthropic Claude Motion Engine, Dashboards Beta, and System Call Containment Policy", "publisher": "Anthropic PBC", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "meta-muse-windows-native", "title": "Meta Muse Desktop Architecture: Native Windows Runtime, Local Memory Buffers, and Cross-Platform Integration", "publisher": "Meta Platforms & Microsoft", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "deepseek-v4", "title": "DeepSeek V4 Preview", "publisher": "DeepSeek", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true }, { "id": "qwen-audio3", "title": "Qwen-Audio-3.0-TTS: More Multilingual, Easier to Direct", "publisher": "Alibaba Cloud / Qwen", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "amd-anthropic", "title": "AMD and Anthropic announce strategic partnership for up to 2 GW of MI450 Series GPUs", "publisher": "AMD Newsroom", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "reuters-aisi-agent-evals-aug5", "title": "OpenAI, Anthropic AI agents implicated in new security breaches", "publisher": "Reuters", "grade": "C", "score": 71, "freshnessStatus": "current", "critical": false }, { "id": "arcprize-opus5-jul24", "title": "Claude Opus 5 — ARC-AGI results", "publisher": "ARC Prize", "grade": "A", "score": 93, "freshnessStatus": "current", "critical": false }, { "id": "arcprize-gpt56-jul9", "title": "GPT-5.6 — ARC-AGI results", "publisher": "ARC Prize", "grade": "A", "score": 93, "freshnessStatus": "current", "critical": false }, { "id": "artificialanalysis-v4flash-aug9", "title": "DeepSeek V4 Flash — independent model analysis", "publisher": "Artificial Analysis", "grade": "C", "score": 71, "freshnessStatus": "current", "critical": false }, { "id": "aisi-test-time-compute", "title": "Measuring AI agents progress on multi-step cyber attack scenarios", "publisher": "UK AI Security Institute", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": false }, { "id": "cvpr-agentsafe-2026", "title": "AGENTSAFE: Benchmarking the Safety of Embodied Agents on Hazardous Instructions", "publisher": "CVPR 2026", "grade": "B", "score": 85, "freshnessStatus": "current", "critical": false }, { "id": "acl-safeagent-2026", "title": "SafeAgent: Safeguarding LLM Agents via an Automated Risk Simulator", "publisher": "ACL 2026", "grade": "B", "score": 85, "freshnessStatus": "current", "critical": false }, { "id": "arxiv-runtime-contract-aug11", "title": "Agent Safety Should Be a Runtime Contract", "publisher": "arXiv", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": false }, { "id": "reuters-zai-glm53-aug14", "title": "China's Z.ai says new model nears Anthropic's Mythos 5 in cyber-defence tests", "publisher": "Reuters", "grade": "A", "score": 93, "freshnessStatus": "current", "critical": false }, { "id": "arxiv-multi-agent-institutions-aug10", "title": "Multi-Agent AI Safety as an Institutional Design Problem", "publisher": "arXiv", "grade": "A", "score": 93, "freshnessStatus": "current", "critical": false }, { "id": "uscc-two-loops-mar23", "title": "Two Loops: How China’s Open AI Strategy Reinforces Its Industrial Dominance", "publisher": "U.S.-China Economic and Security Review Commission", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": false }, { "id": "reuters-unitree-us-research-aug18", "title": "How US military funding propelled China’s robot dogs", "publisher": "Reuters", "grade": "A", "score": 93, "freshnessStatus": "current", "critical": false }, { "id": "openai-third-party-cyber-aug4", "title": "Third-party cyber evaluations involving OpenAI models", "publisher": "OpenAI", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": false }, { "id": "jll-emea-dc-aug3", "title": "EMEA data centre mid-year 2026 report", "publisher": "JLL", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": false }, { "id": "reuters-aisi-agent-aug20", "title": "How a Texas student blew the whistle on a rogue AI hacking attempt", "publisher": "Reuters", "grade": "C", "score": 71, "freshnessStatus": "current", "critical": false }, { "id": "arc-anthropic-cot-faithfulness", "title": "Evaluating Chain-of-Thought Faithfulness Under Outcome Reinforcement Learning", "publisher": "Anthropic Research & Alignment Research Center", "grade": "A", "score": 93, "freshnessStatus": "current", "critical": false }, { "id": "llama-3-4-moe-release", "title": "Meta Llama 3.4-MoE Weights Release & Speculative Draft Heads", "publisher": "Meta AI / Hugging Face", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": false }, { "id": "stanford-berkeley-steering", "title": "Deterministic Containment of Autonomous Agent Tool Invocation via Residual Activation Steering", "publisher": "Stanford CRFM & UC Berkeley AI Research", "grade": "A", "score": 93, "freshnessStatus": "current", "critical": false }, { "id": "bird-sql-evaluation-harness", "title": "BIRD-SQL Cross-Domain Enterprise Relational Database Benchmark Leaderboard", "publisher": "BIRD-SQL Consortium / DAMO", "grade": "A", "score": 93, "freshnessStatus": "current", "critical": false }, { "id": "kv-cache-mla-systems", "title": "Multi-Head Latent Attention and 4-Bit PagedAttention Serving Implementations", "publisher": "vLLM Project & TensorRT-LLM", "grade": "A", "score": 93, "freshnessStatus": "current", "critical": false }, { "id": "mit-cmu-decomposed-semantic-inversion", "title": "Decomposed Semantic Inversion in Long-Context LLMs", "publisher": "MIT CSAIL & Carnegie Mellon University", "grade": "A", "score": 93, "freshnessStatus": "current", "critical": false }, { "id": "swebench-pro-evaluation", "title": "SWE-bench Verified Saturation Analysis and SWE-bench Pro Evaluation Methodology", "publisher": "SWE-bench Consortium", "grade": "A", "score": 93, "freshnessStatus": "current", "critical": false }, { "id": "nvidia-rubin-nvl72-infrastructure", "title": "NVIDIA Rubin NVL72 and Blackwell Ultra Rack-Scale Architecture Specifications", "publisher": "NVIDIA Corporation", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": false }, { "id": "project-glasswing-safeguards", "title": "Project Glasswing Defensive Cybersecurity Verification Framework and Enterprise Frontier Safeguards", "publisher": "Anthropic & Coalition Partners", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": false }, { "id": "cmu-berkeley-agent-injection-audit", "title": "Indirect Prompt Injection and Tool Poisoning in Autonomous Multi-Turn Agents: Empirical Vulnerability Assessment and Structural Defenses", "publisher": "Carnegie Mellon University & UC Berkeley", "grade": "A", "score": 93, "freshnessStatus": "current", "critical": false }, { "id": "osworld-webarena-benchmark-audit", "title": "OSWorld and WebArena Autonomous Agent Evaluation Leaderboards and Dynamic Environment Test Logs", "publisher": "OSWorld / WebArena Evaluation Consortium", "grade": "A", "score": 93, "freshnessStatus": "current", "critical": false }, { "id": "eth-berkeley-speculative-leak-paper", "title": "Speculative Leak: Microarchitectural Side-Channels and Cache Contention in Speculative Decoding Inference Engines (CVE-2026-38192)", "publisher": "ETH Zurich & UC Berkeley Systems and Security Lab", "grade": "A", "score": 93, "freshnessStatus": "current", "critical": false }, { "id": "swe-bench-verified-diagnostic-audit", "title": "Diagnostic Failure Modes of Autonomous Coding Agents on SWE-bench Verified: The State Synchronization Bottleneck", "publisher": "Open-Source Software Engineering Evaluation Collective", "grade": "A", "score": 93, "freshnessStatus": "current", "critical": false }, { "id": "bair-stanford-quantization-safety-drift", "title": "Asymmetric Representation Degradation: Why Extreme Quantization Erodes Safety Alignment Faster Than General Intelligence", "publisher": "Berkeley AI Research (BAIR) & Stanford Center for Research on Foundation Models (CRFM)", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": false }, { "id": "google-gemini-2-5-pro-reasoner-paper", "title": "Gemini 2.5 Pro Reasoner Technical Report: Step-Level Supervision, Monte Carlo Tree Search, and Robotics Foundation API", "publisher": "Google DeepMind", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "mistral-large-3-hybrid-ssm-paper", "title": "Mistral Large 3: Dense 145B Hybrid Selective State-Space and Transformer Model Architecture", "publisher": "Mistral AI Research", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "anthropic-multimodal-prompt-smuggling", "title": "Multi-Modal Boundary Breaches in Autonomous Execution Stacks: Exploiting Visual Representation Layers for Agent Privilege Escalation", "publisher": "Anthropic Alignment & Red-Teaming Research", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": false }, { "id": "aisi-joint-multimodal-agent-sandbox-audit", "title": "US and UK AI Safety Institutes: Joint Empirical Audit on Multimodal Visual Prompt Smuggling and Mandatory Sandbox Virtualization for Desktop Agents", "publisher": "US AI Safety Institute (NIST) / UK AI Safety Institute (DSIT)", "grade": "C", "score": 71, "freshnessStatus": "current", "critical": true }, { "id": "meta-llama-4-speculative-decoding-preview", "title": "Meta AI Engineering: Llama 4 Architectural Blueprint, Native Speculative Decoding, and MoE Routing at Scale", "publisher": "Meta AI / Fundamental AI Research", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "nvidia-tensorrt-llm-fp4-blackwell", "title": "NVIDIA TensorRT-LLM 0.16 General Availability: FP4 Microscaling Acceleration on Blackwell GB200 Systems", "publisher": "NVIDIA Technical Blog", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": false }, { "id": "nvidia-blackwell-ultra-nvl72-fp4", "title": "NVIDIA Blackwell Ultra GB200 NVL72 System Architecture: 3.2x Multi-Turn Chain-of-Thought Inference via FP4 Microscaling", "publisher": "NVIDIA Technical Blog", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "deepseek-ascend-cann9-cluster-whitepaper", "title": "DeepSeek 160,000-Accelerator Megacluster on Huawei Ascend CANN 9.0 Architecture", "publisher": "DeepSeek AI / Industry Filings", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true } ], "bestGrade": "A", "worstGrade": "C", "coverage": { "status": "supported", "reason": "At least one valid source explicitly supports this claim." } }, { "id": "claim:benchmark.items.5", "path": "benchmark.items.5", "section": "benchmarks", "kind": "numeric-record", "critical": true, "text": "GPT-5.6 Luna — score: 51.2", "sourceIds": [ "openai-gpt56", "anthropic-claude-3-7-sonnet-system-card", "openai-o3-pro-system-card", "xai-grok45", "moonshot-kimi-k3", "zhipu-glm52", "alibaba-qwen37", "alibaba-qwen-2-5-coder-reasoner-paper", "swe-bench-verified-october-audit", "google-gemini35", "zhipu-glm-ocr", "baidu-ernie50", "openai-operator-sandbox-architecture", "nist-aisi-agentic-privilege-escalation-bench", "anthropic-claude-3-8-system-card", "deepseek-v3-5-architecture-report", "anthropic-claude-motion-usage-policy", "meta-muse-windows-native", "deepseek-v4", "qwen-audio3", "amd-anthropic", "reuters-aisi-agent-evals-aug5", "arcprize-opus5-jul24", "arcprize-gpt56-jul9", "artificialanalysis-v4flash-aug9", "aisi-test-time-compute", "cvpr-agentsafe-2026", "acl-safeagent-2026", "arxiv-runtime-contract-aug11", "reuters-zai-glm53-aug14", "arxiv-multi-agent-institutions-aug10", "uscc-two-loops-mar23", "reuters-unitree-us-research-aug18", "openai-third-party-cyber-aug4", "jll-emea-dc-aug3", "reuters-aisi-agent-aug20", "arc-anthropic-cot-faithfulness", "llama-3-4-moe-release", "stanford-berkeley-steering", "bird-sql-evaluation-harness", "kv-cache-mla-systems", "mit-cmu-decomposed-semantic-inversion", "swebench-pro-evaluation", "nvidia-rubin-nvl72-infrastructure", "project-glasswing-safeguards", "cmu-berkeley-agent-injection-audit", "osworld-webarena-benchmark-audit", "eth-berkeley-speculative-leak-paper", "swe-bench-verified-diagnostic-audit", "bair-stanford-quantization-safety-drift", "google-gemini-2-5-pro-reasoner-paper", "mistral-large-3-hybrid-ssm-paper", "anthropic-multimodal-prompt-smuggling", "aisi-joint-multimodal-agent-sandbox-audit", "meta-llama-4-speculative-decoding-preview", "nvidia-tensorrt-llm-fp4-blackwell", "nvidia-blackwell-ultra-nvl72-fp4", "deepseek-ascend-cann9-cluster-whitepaper" ], "citationOrigin": "explicit-claim-map", "evidenceMode": "registry-review", "evidenceNote": null, "numericValues": [ "51.2" ], "isNumeric": true, "citationCount": 58, "validSourceIds": [ "openai-gpt56", "anthropic-claude-3-7-sonnet-system-card", "openai-o3-pro-system-card", "xai-grok45", "moonshot-kimi-k3", "zhipu-glm52", "alibaba-qwen37", "alibaba-qwen-2-5-coder-reasoner-paper", "swe-bench-verified-october-audit", "google-gemini35", "zhipu-glm-ocr", "baidu-ernie50", "openai-operator-sandbox-architecture", "nist-aisi-agentic-privilege-escalation-bench", "anthropic-claude-3-8-system-card", "deepseek-v3-5-architecture-report", "anthropic-claude-motion-usage-policy", "meta-muse-windows-native", "deepseek-v4", "qwen-audio3", "amd-anthropic", "reuters-aisi-agent-evals-aug5", "arcprize-opus5-jul24", "arcprize-gpt56-jul9", "artificialanalysis-v4flash-aug9", "aisi-test-time-compute", "cvpr-agentsafe-2026", "acl-safeagent-2026", "arxiv-runtime-contract-aug11", "reuters-zai-glm53-aug14", "arxiv-multi-agent-institutions-aug10", "uscc-two-loops-mar23", "reuters-unitree-us-research-aug18", "openai-third-party-cyber-aug4", "jll-emea-dc-aug3", "reuters-aisi-agent-aug20", "arc-anthropic-cot-faithfulness", "llama-3-4-moe-release", "stanford-berkeley-steering", "bird-sql-evaluation-harness", "kv-cache-mla-systems", "mit-cmu-decomposed-semantic-inversion", "swebench-pro-evaluation", "nvidia-rubin-nvl72-infrastructure", "project-glasswing-safeguards", "cmu-berkeley-agent-injection-audit", "osworld-webarena-benchmark-audit", "eth-berkeley-speculative-leak-paper", "swe-bench-verified-diagnostic-audit", "bair-stanford-quantization-safety-drift", "google-gemini-2-5-pro-reasoner-paper", "mistral-large-3-hybrid-ssm-paper", "anthropic-multimodal-prompt-smuggling", "aisi-joint-multimodal-agent-sandbox-audit", "meta-llama-4-speculative-decoding-preview", "nvidia-tensorrt-llm-fp4-blackwell", "nvidia-blackwell-ultra-nvl72-fp4", "deepseek-ascend-cann9-cluster-whitepaper" ], "invalidSourceIds": [], "sourceDetails": [ { "id": "openai-gpt56", "title": "Introducing GPT-5.6", "publisher": "OpenAI", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true }, { "id": "anthropic-claude-3-7-sonnet-system-card", "title": "Anthropic Claude 3.7 Sonnet: Hybrid Reasoning Architecture, Extended Thinking Token Scaling, and Sub-Pixel Computer Use v2", "publisher": "Anthropic", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "openai-o3-pro-system-card", "title": "OpenAI o3-pro System Card: Frontier Reasoning, Test-Time Tree Search with Native Multi-Step Tool Execution, and Autonomous Coding Evaluations", "publisher": "OpenAI", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "xai-grok45", "title": "Grok 4.5", "publisher": "xAI", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true }, { "id": "moonshot-kimi-k3", "title": "Kimi API platform — Kimi K3 model and pricing", "publisher": "Moonshot AI", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true }, { "id": "zhipu-glm52", "title": "Zhipu BigModel — GLM-5.2 model overview", "publisher": "Zhipu AI", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true }, { "id": "alibaba-qwen37", "title": "Alibaba Cloud Model Studio — Qwen flagship models", "publisher": "Alibaba Cloud", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true }, { "id": "alibaba-qwen-2-5-coder-reasoner-paper", "title": "Qwen 2.5-Coder-32B-Reasoner: Open-Weight Test-Time Search Code Agent with Step-Level Process Reward Supervision", "publisher": "Alibaba Cloud & Qwen Team", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "swe-bench-verified-october-audit", "title": "SWE-bench Verified & Terminal-Bench October 2026 Frontier Coding Leaderboard: Test-Time Reasoning vs. Open-Weight Architectures", "publisher": "CodeEval Foundation / SWE-bench Consortium", "grade": "A", "score": 97, "freshnessStatus": "current", "critical": true }, { "id": "google-gemini35", "title": "Gemini 3.5", "publisher": "Google", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true }, { "id": "zhipu-glm-ocr", "title": "Zhipu BigModel — GLM-OCR", "publisher": "Zhipu AI", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "baidu-ernie50", "title": "Baidu Qianfan — ERNIE 5.0 model list and international price", "publisher": "Baidu AI Cloud", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true }, { "id": "openai-operator-sandbox-architecture", "title": "OpenAI Operator Architecture: MicroVM Hardware Isolation, Permission Gating, and Dual-Loop Verification for Web Agents", "publisher": "OpenAI", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "nist-aisi-agentic-privilege-escalation-bench", "title": "US Artificial Intelligence Safety Institute (NIST AISI): Standardized Test Suite for Agentic Privilege Escalation and MicroVM Containment", "publisher": "National Institute of Standards and Technology (NIST)", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "anthropic-claude-3-8-system-card", "title": "Anthropic Claude 3.8 Architecture & System Card: Computer Use v2.5, POSIX Bash Sandboxing, and SWE-bench Verified Benchmark", "publisher": "Anthropic PBC", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "deepseek-v3-5-architecture-report", "title": "DeepSeek-V3.5 Technical Report: Multi-Head Latent Attention (MLA) Scaling and DualPipe Parallelism across 2,048 Compute Nodes", "publisher": "DeepSeek AI", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "anthropic-claude-motion-usage-policy", "title": "Anthropic Claude Motion Engine, Dashboards Beta, and System Call Containment Policy", "publisher": "Anthropic PBC", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "meta-muse-windows-native", "title": "Meta Muse Desktop Architecture: Native Windows Runtime, Local Memory Buffers, and Cross-Platform Integration", "publisher": "Meta Platforms & Microsoft", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "deepseek-v4", "title": "DeepSeek V4 Preview", "publisher": "DeepSeek", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true }, { "id": "qwen-audio3", "title": "Qwen-Audio-3.0-TTS: More Multilingual, Easier to Direct", "publisher": "Alibaba Cloud / Qwen", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "amd-anthropic", "title": "AMD and Anthropic announce strategic partnership for up to 2 GW of MI450 Series GPUs", "publisher": "AMD Newsroom", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "reuters-aisi-agent-evals-aug5", "title": "OpenAI, Anthropic AI agents implicated in new security breaches", "publisher": "Reuters", "grade": "C", "score": 71, "freshnessStatus": "current", "critical": false }, { "id": "arcprize-opus5-jul24", "title": "Claude Opus 5 — ARC-AGI results", "publisher": "ARC Prize", "grade": "A", "score": 93, "freshnessStatus": "current", "critical": false }, { "id": "arcprize-gpt56-jul9", "title": "GPT-5.6 — ARC-AGI results", "publisher": "ARC Prize", "grade": "A", "score": 93, "freshnessStatus": "current", "critical": false }, { "id": "artificialanalysis-v4flash-aug9", "title": "DeepSeek V4 Flash — independent model analysis", "publisher": "Artificial Analysis", "grade": "C", "score": 71, "freshnessStatus": "current", "critical": false }, { "id": "aisi-test-time-compute", "title": "Measuring AI agents progress on multi-step cyber attack scenarios", "publisher": "UK AI Security Institute", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": false }, { "id": "cvpr-agentsafe-2026", "title": "AGENTSAFE: Benchmarking the Safety of Embodied Agents on Hazardous Instructions", "publisher": "CVPR 2026", "grade": "B", "score": 85, "freshnessStatus": "current", "critical": false }, { "id": "acl-safeagent-2026", "title": "SafeAgent: Safeguarding LLM Agents via an Automated Risk Simulator", "publisher": "ACL 2026", "grade": "B", "score": 85, "freshnessStatus": "current", "critical": false }, { "id": "arxiv-runtime-contract-aug11", "title": "Agent Safety Should Be a Runtime Contract", "publisher": "arXiv", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": false }, { "id": "reuters-zai-glm53-aug14", "title": "China's Z.ai says new model nears Anthropic's Mythos 5 in cyber-defence tests", "publisher": "Reuters", "grade": "A", "score": 93, "freshnessStatus": "current", "critical": false }, { "id": "arxiv-multi-agent-institutions-aug10", "title": "Multi-Agent AI Safety as an Institutional Design Problem", "publisher": "arXiv", "grade": "A", "score": 93, "freshnessStatus": "current", "critical": false }, { "id": "uscc-two-loops-mar23", "title": "Two Loops: How China’s Open AI Strategy Reinforces Its Industrial Dominance", "publisher": "U.S.-China Economic and Security Review Commission", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": false }, { "id": "reuters-unitree-us-research-aug18", "title": "How US military funding propelled China’s robot dogs", "publisher": "Reuters", "grade": "A", "score": 93, "freshnessStatus": "current", "critical": false }, { "id": "openai-third-party-cyber-aug4", "title": "Third-party cyber evaluations involving OpenAI models", "publisher": "OpenAI", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": false }, { "id": "jll-emea-dc-aug3", "title": "EMEA data centre mid-year 2026 report", "publisher": "JLL", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": false }, { "id": "reuters-aisi-agent-aug20", "title": "How a Texas student blew the whistle on a rogue AI hacking attempt", "publisher": "Reuters", "grade": "C", "score": 71, "freshnessStatus": "current", "critical": false }, { "id": "arc-anthropic-cot-faithfulness", "title": "Evaluating Chain-of-Thought Faithfulness Under Outcome Reinforcement Learning", "publisher": "Anthropic Research & Alignment Research Center", "grade": "A", "score": 93, "freshnessStatus": "current", "critical": false }, { "id": "llama-3-4-moe-release", "title": "Meta Llama 3.4-MoE Weights Release & Speculative Draft Heads", "publisher": "Meta AI / Hugging Face", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": false }, { "id": "stanford-berkeley-steering", "title": "Deterministic Containment of Autonomous Agent Tool Invocation via Residual Activation Steering", "publisher": "Stanford CRFM & UC Berkeley AI Research", "grade": "A", "score": 93, "freshnessStatus": "current", "critical": false }, { "id": "bird-sql-evaluation-harness", "title": "BIRD-SQL Cross-Domain Enterprise Relational Database Benchmark Leaderboard", "publisher": "BIRD-SQL Consortium / DAMO", "grade": "A", "score": 93, "freshnessStatus": "current", "critical": false }, { "id": "kv-cache-mla-systems", "title": "Multi-Head Latent Attention and 4-Bit PagedAttention Serving Implementations", "publisher": "vLLM Project & TensorRT-LLM", "grade": "A", "score": 93, "freshnessStatus": "current", "critical": false }, { "id": "mit-cmu-decomposed-semantic-inversion", "title": "Decomposed Semantic Inversion in Long-Context LLMs", "publisher": "MIT CSAIL & Carnegie Mellon University", "grade": "A", "score": 93, "freshnessStatus": "current", "critical": false }, { "id": "swebench-pro-evaluation", "title": "SWE-bench Verified Saturation Analysis and SWE-bench Pro Evaluation Methodology", "publisher": "SWE-bench Consortium", "grade": "A", "score": 93, "freshnessStatus": "current", "critical": false }, { "id": "nvidia-rubin-nvl72-infrastructure", "title": "NVIDIA Rubin NVL72 and Blackwell Ultra Rack-Scale Architecture Specifications", "publisher": "NVIDIA Corporation", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": false }, { "id": "project-glasswing-safeguards", "title": "Project Glasswing Defensive Cybersecurity Verification Framework and Enterprise Frontier Safeguards", "publisher": "Anthropic & Coalition Partners", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": false }, { "id": "cmu-berkeley-agent-injection-audit", "title": "Indirect Prompt Injection and Tool Poisoning in Autonomous Multi-Turn Agents: Empirical Vulnerability Assessment and Structural Defenses", "publisher": "Carnegie Mellon University & UC Berkeley", "grade": "A", "score": 93, "freshnessStatus": "current", "critical": false }, { "id": "osworld-webarena-benchmark-audit", "title": "OSWorld and WebArena Autonomous Agent Evaluation Leaderboards and Dynamic Environment Test Logs", "publisher": "OSWorld / WebArena Evaluation Consortium", "grade": "A", "score": 93, "freshnessStatus": "current", "critical": false }, { "id": "eth-berkeley-speculative-leak-paper", "title": "Speculative Leak: Microarchitectural Side-Channels and Cache Contention in Speculative Decoding Inference Engines (CVE-2026-38192)", "publisher": "ETH Zurich & UC Berkeley Systems and Security Lab", "grade": "A", "score": 93, "freshnessStatus": "current", "critical": false }, { "id": "swe-bench-verified-diagnostic-audit", "title": "Diagnostic Failure Modes of Autonomous Coding Agents on SWE-bench Verified: The State Synchronization Bottleneck", "publisher": "Open-Source Software Engineering Evaluation Collective", "grade": "A", "score": 93, "freshnessStatus": "current", "critical": false }, { "id": "bair-stanford-quantization-safety-drift", "title": "Asymmetric Representation Degradation: Why Extreme Quantization Erodes Safety Alignment Faster Than General Intelligence", "publisher": "Berkeley AI Research (BAIR) & Stanford Center for Research on Foundation Models (CRFM)", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": false }, { "id": "google-gemini-2-5-pro-reasoner-paper", "title": "Gemini 2.5 Pro Reasoner Technical Report: Step-Level Supervision, Monte Carlo Tree Search, and Robotics Foundation API", "publisher": "Google DeepMind", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "mistral-large-3-hybrid-ssm-paper", "title": "Mistral Large 3: Dense 145B Hybrid Selective State-Space and Transformer Model Architecture", "publisher": "Mistral AI Research", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "anthropic-multimodal-prompt-smuggling", "title": "Multi-Modal Boundary Breaches in Autonomous Execution Stacks: Exploiting Visual Representation Layers for Agent Privilege Escalation", "publisher": "Anthropic Alignment & Red-Teaming Research", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": false }, { "id": "aisi-joint-multimodal-agent-sandbox-audit", "title": "US and UK AI Safety Institutes: Joint Empirical Audit on Multimodal Visual Prompt Smuggling and Mandatory Sandbox Virtualization for Desktop Agents", "publisher": "US AI Safety Institute (NIST) / UK AI Safety Institute (DSIT)", "grade": "C", "score": 71, "freshnessStatus": "current", "critical": true }, { "id": "meta-llama-4-speculative-decoding-preview", "title": "Meta AI Engineering: Llama 4 Architectural Blueprint, Native Speculative Decoding, and MoE Routing at Scale", "publisher": "Meta AI / Fundamental AI Research", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "nvidia-tensorrt-llm-fp4-blackwell", "title": "NVIDIA TensorRT-LLM 0.16 General Availability: FP4 Microscaling Acceleration on Blackwell GB200 Systems", "publisher": "NVIDIA Technical Blog", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": false }, { "id": "nvidia-blackwell-ultra-nvl72-fp4", "title": "NVIDIA Blackwell Ultra GB200 NVL72 System Architecture: 3.2x Multi-Turn Chain-of-Thought Inference via FP4 Microscaling", "publisher": "NVIDIA Technical Blog", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "deepseek-ascend-cann9-cluster-whitepaper", "title": "DeepSeek 160,000-Accelerator Megacluster on Huawei Ascend CANN 9.0 Architecture", "publisher": "DeepSeek AI / Industry Filings", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true } ], "bestGrade": "A", "worstGrade": "C", "coverage": { "status": "supported", "reason": "At least one valid source explicitly supports this claim." } }, { "id": "claim:benchmark.items.6", "path": "benchmark.items.6", "section": "benchmarks", "kind": "numeric-record", "critical": true, "text": "Gemini 3.5 Flash — score: 50.2", "sourceIds": [ "openai-gpt56", "anthropic-claude-3-7-sonnet-system-card", "openai-o3-pro-system-card", "xai-grok45", "moonshot-kimi-k3", "zhipu-glm52", "alibaba-qwen37", "alibaba-qwen-2-5-coder-reasoner-paper", "swe-bench-verified-october-audit", "google-gemini35", "zhipu-glm-ocr", "baidu-ernie50", "openai-operator-sandbox-architecture", "nist-aisi-agentic-privilege-escalation-bench", "anthropic-claude-3-8-system-card", "deepseek-v3-5-architecture-report", "anthropic-claude-motion-usage-policy", "meta-muse-windows-native", "deepseek-v4", "qwen-audio3", "amd-anthropic", "reuters-aisi-agent-evals-aug5", "arcprize-opus5-jul24", "arcprize-gpt56-jul9", "artificialanalysis-v4flash-aug9", "aisi-test-time-compute", "cvpr-agentsafe-2026", "acl-safeagent-2026", "arxiv-runtime-contract-aug11", "reuters-zai-glm53-aug14", "arxiv-multi-agent-institutions-aug10", "uscc-two-loops-mar23", "reuters-unitree-us-research-aug18", "openai-third-party-cyber-aug4", "jll-emea-dc-aug3", "reuters-aisi-agent-aug20", "arc-anthropic-cot-faithfulness", "llama-3-4-moe-release", "stanford-berkeley-steering", "bird-sql-evaluation-harness", "kv-cache-mla-systems", "mit-cmu-decomposed-semantic-inversion", "swebench-pro-evaluation", "nvidia-rubin-nvl72-infrastructure", "project-glasswing-safeguards", "cmu-berkeley-agent-injection-audit", "osworld-webarena-benchmark-audit", "eth-berkeley-speculative-leak-paper", "swe-bench-verified-diagnostic-audit", "bair-stanford-quantization-safety-drift", "google-gemini-2-5-pro-reasoner-paper", "mistral-large-3-hybrid-ssm-paper", "anthropic-multimodal-prompt-smuggling", "aisi-joint-multimodal-agent-sandbox-audit", "meta-llama-4-speculative-decoding-preview", "nvidia-tensorrt-llm-fp4-blackwell", "nvidia-blackwell-ultra-nvl72-fp4", "deepseek-ascend-cann9-cluster-whitepaper" ], "citationOrigin": "explicit-claim-map", "evidenceMode": "registry-review", "evidenceNote": null, "numericValues": [ "50.2" ], "isNumeric": true, "citationCount": 58, "validSourceIds": [ "openai-gpt56", "anthropic-claude-3-7-sonnet-system-card", "openai-o3-pro-system-card", "xai-grok45", "moonshot-kimi-k3", "zhipu-glm52", "alibaba-qwen37", "alibaba-qwen-2-5-coder-reasoner-paper", "swe-bench-verified-october-audit", "google-gemini35", "zhipu-glm-ocr", "baidu-ernie50", "openai-operator-sandbox-architecture", "nist-aisi-agentic-privilege-escalation-bench", "anthropic-claude-3-8-system-card", "deepseek-v3-5-architecture-report", "anthropic-claude-motion-usage-policy", "meta-muse-windows-native", "deepseek-v4", "qwen-audio3", "amd-anthropic", "reuters-aisi-agent-evals-aug5", "arcprize-opus5-jul24", "arcprize-gpt56-jul9", "artificialanalysis-v4flash-aug9", "aisi-test-time-compute", "cvpr-agentsafe-2026", "acl-safeagent-2026", "arxiv-runtime-contract-aug11", "reuters-zai-glm53-aug14", "arxiv-multi-agent-institutions-aug10", "uscc-two-loops-mar23", "reuters-unitree-us-research-aug18", "openai-third-party-cyber-aug4", "jll-emea-dc-aug3", "reuters-aisi-agent-aug20", "arc-anthropic-cot-faithfulness", "llama-3-4-moe-release", "stanford-berkeley-steering", "bird-sql-evaluation-harness", "kv-cache-mla-systems", "mit-cmu-decomposed-semantic-inversion", "swebench-pro-evaluation", "nvidia-rubin-nvl72-infrastructure", "project-glasswing-safeguards", "cmu-berkeley-agent-injection-audit", "osworld-webarena-benchmark-audit", "eth-berkeley-speculative-leak-paper", "swe-bench-verified-diagnostic-audit", "bair-stanford-quantization-safety-drift", "google-gemini-2-5-pro-reasoner-paper", "mistral-large-3-hybrid-ssm-paper", "anthropic-multimodal-prompt-smuggling", "aisi-joint-multimodal-agent-sandbox-audit", "meta-llama-4-speculative-decoding-preview", "nvidia-tensorrt-llm-fp4-blackwell", "nvidia-blackwell-ultra-nvl72-fp4", "deepseek-ascend-cann9-cluster-whitepaper" ], "invalidSourceIds": [], "sourceDetails": [ { "id": "openai-gpt56", "title": "Introducing GPT-5.6", "publisher": "OpenAI", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true }, { "id": "anthropic-claude-3-7-sonnet-system-card", "title": "Anthropic Claude 3.7 Sonnet: Hybrid Reasoning Architecture, Extended Thinking Token Scaling, and Sub-Pixel Computer Use v2", "publisher": "Anthropic", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "openai-o3-pro-system-card", "title": "OpenAI o3-pro System Card: Frontier Reasoning, Test-Time Tree Search with Native Multi-Step Tool Execution, and Autonomous Coding Evaluations", "publisher": "OpenAI", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "xai-grok45", "title": "Grok 4.5", "publisher": "xAI", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true }, { "id": "moonshot-kimi-k3", "title": "Kimi API platform — Kimi K3 model and pricing", "publisher": "Moonshot AI", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true }, { "id": "zhipu-glm52", "title": "Zhipu BigModel — GLM-5.2 model overview", "publisher": "Zhipu AI", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true }, { "id": "alibaba-qwen37", "title": "Alibaba Cloud Model Studio — Qwen flagship models", "publisher": "Alibaba Cloud", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true }, { "id": "alibaba-qwen-2-5-coder-reasoner-paper", "title": "Qwen 2.5-Coder-32B-Reasoner: Open-Weight Test-Time Search Code Agent with Step-Level Process Reward Supervision", "publisher": "Alibaba Cloud & Qwen Team", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "swe-bench-verified-october-audit", "title": "SWE-bench Verified & Terminal-Bench October 2026 Frontier Coding Leaderboard: Test-Time Reasoning vs. Open-Weight Architectures", "publisher": "CodeEval Foundation / SWE-bench Consortium", "grade": "A", "score": 97, "freshnessStatus": "current", "critical": true }, { "id": "google-gemini35", "title": "Gemini 3.5", "publisher": "Google", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true }, { "id": "zhipu-glm-ocr", "title": "Zhipu BigModel — GLM-OCR", "publisher": "Zhipu AI", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "baidu-ernie50", "title": "Baidu Qianfan — ERNIE 5.0 model list and international price", "publisher": "Baidu AI Cloud", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true }, { "id": "openai-operator-sandbox-architecture", "title": "OpenAI Operator Architecture: MicroVM Hardware Isolation, Permission Gating, and Dual-Loop Verification for Web Agents", "publisher": "OpenAI", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "nist-aisi-agentic-privilege-escalation-bench", "title": "US Artificial Intelligence Safety Institute (NIST AISI): Standardized Test Suite for Agentic Privilege Escalation and MicroVM Containment", "publisher": "National Institute of Standards and Technology (NIST)", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "anthropic-claude-3-8-system-card", "title": "Anthropic Claude 3.8 Architecture & System Card: Computer Use v2.5, POSIX Bash Sandboxing, and SWE-bench Verified Benchmark", "publisher": "Anthropic PBC", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "deepseek-v3-5-architecture-report", "title": "DeepSeek-V3.5 Technical Report: Multi-Head Latent Attention (MLA) Scaling and DualPipe Parallelism across 2,048 Compute Nodes", "publisher": "DeepSeek AI", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "anthropic-claude-motion-usage-policy", "title": "Anthropic Claude Motion Engine, Dashboards Beta, and System Call Containment Policy", "publisher": "Anthropic PBC", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "meta-muse-windows-native", "title": "Meta Muse Desktop Architecture: Native Windows Runtime, Local Memory Buffers, and Cross-Platform Integration", "publisher": "Meta Platforms & Microsoft", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "deepseek-v4", "title": "DeepSeek V4 Preview", "publisher": "DeepSeek", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true }, { "id": "qwen-audio3", "title": "Qwen-Audio-3.0-TTS: More Multilingual, Easier to Direct", "publisher": "Alibaba Cloud / Qwen", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "amd-anthropic", "title": "AMD and Anthropic announce strategic partnership for up to 2 GW of MI450 Series GPUs", "publisher": "AMD Newsroom", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "reuters-aisi-agent-evals-aug5", "title": "OpenAI, Anthropic AI agents implicated in new security breaches", "publisher": "Reuters", "grade": "C", "score": 71, "freshnessStatus": "current", "critical": false }, { "id": "arcprize-opus5-jul24", "title": "Claude Opus 5 — ARC-AGI results", "publisher": "ARC Prize", "grade": "A", "score": 93, "freshnessStatus": "current", "critical": false }, { "id": "arcprize-gpt56-jul9", "title": "GPT-5.6 — ARC-AGI results", "publisher": "ARC Prize", "grade": "A", "score": 93, "freshnessStatus": "current", "critical": false }, { "id": "artificialanalysis-v4flash-aug9", "title": "DeepSeek V4 Flash — independent model analysis", "publisher": "Artificial Analysis", "grade": "C", "score": 71, "freshnessStatus": "current", "critical": false }, { "id": "aisi-test-time-compute", "title": "Measuring AI agents progress on multi-step cyber attack scenarios", "publisher": "UK AI Security Institute", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": false }, { "id": "cvpr-agentsafe-2026", "title": "AGENTSAFE: Benchmarking the Safety of Embodied Agents on Hazardous Instructions", "publisher": "CVPR 2026", "grade": "B", "score": 85, "freshnessStatus": "current", "critical": false }, { "id": "acl-safeagent-2026", "title": "SafeAgent: Safeguarding LLM Agents via an Automated Risk Simulator", "publisher": "ACL 2026", "grade": "B", "score": 85, "freshnessStatus": "current", "critical": false }, { "id": "arxiv-runtime-contract-aug11", "title": "Agent Safety Should Be a Runtime Contract", "publisher": "arXiv", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": false }, { "id": "reuters-zai-glm53-aug14", "title": "China's Z.ai says new model nears Anthropic's Mythos 5 in cyber-defence tests", "publisher": "Reuters", "grade": "A", "score": 93, "freshnessStatus": "current", "critical": false }, { "id": "arxiv-multi-agent-institutions-aug10", "title": "Multi-Agent AI Safety as an Institutional Design Problem", "publisher": "arXiv", "grade": "A", "score": 93, "freshnessStatus": "current", "critical": false }, { "id": "uscc-two-loops-mar23", "title": "Two Loops: How China’s Open AI Strategy Reinforces Its Industrial Dominance", "publisher": "U.S.-China Economic and Security Review Commission", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": false }, { "id": "reuters-unitree-us-research-aug18", "title": "How US military funding propelled China’s robot dogs", "publisher": "Reuters", "grade": "A", "score": 93, "freshnessStatus": "current", "critical": false }, { "id": "openai-third-party-cyber-aug4", "title": "Third-party cyber evaluations involving OpenAI models", "publisher": "OpenAI", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": false }, { "id": "jll-emea-dc-aug3", "title": "EMEA data centre mid-year 2026 report", "publisher": "JLL", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": false }, { "id": "reuters-aisi-agent-aug20", "title": "How a Texas student blew the whistle on a rogue AI hacking attempt", "publisher": "Reuters", "grade": "C", "score": 71, "freshnessStatus": "current", "critical": false }, { "id": "arc-anthropic-cot-faithfulness", "title": "Evaluating Chain-of-Thought Faithfulness Under Outcome Reinforcement Learning", "publisher": "Anthropic Research & Alignment Research Center", "grade": "A", "score": 93, "freshnessStatus": "current", "critical": false }, { "id": "llama-3-4-moe-release", "title": "Meta Llama 3.4-MoE Weights Release & Speculative Draft Heads", "publisher": "Meta AI / Hugging Face", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": false }, { "id": "stanford-berkeley-steering", "title": "Deterministic Containment of Autonomous Agent Tool Invocation via Residual Activation Steering", "publisher": "Stanford CRFM & UC Berkeley AI Research", "grade": "A", "score": 93, "freshnessStatus": "current", "critical": false }, { "id": "bird-sql-evaluation-harness", "title": "BIRD-SQL Cross-Domain Enterprise Relational Database Benchmark Leaderboard", "publisher": "BIRD-SQL Consortium / DAMO", "grade": "A", "score": 93, "freshnessStatus": "current", "critical": false }, { "id": "kv-cache-mla-systems", "title": "Multi-Head Latent Attention and 4-Bit PagedAttention Serving Implementations", "publisher": "vLLM Project & TensorRT-LLM", "grade": "A", "score": 93, "freshnessStatus": "current", "critical": false }, { "id": "mit-cmu-decomposed-semantic-inversion", "title": "Decomposed Semantic Inversion in Long-Context LLMs", "publisher": "MIT CSAIL & Carnegie Mellon University", "grade": "A", "score": 93, "freshnessStatus": "current", "critical": false }, { "id": "swebench-pro-evaluation", "title": "SWE-bench Verified Saturation Analysis and SWE-bench Pro Evaluation Methodology", "publisher": "SWE-bench Consortium", "grade": "A", "score": 93, "freshnessStatus": "current", "critical": false }, { "id": "nvidia-rubin-nvl72-infrastructure", "title": "NVIDIA Rubin NVL72 and Blackwell Ultra Rack-Scale Architecture Specifications", "publisher": "NVIDIA Corporation", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": false }, { "id": "project-glasswing-safeguards", "title": "Project Glasswing Defensive Cybersecurity Verification Framework and Enterprise Frontier Safeguards", "publisher": "Anthropic & Coalition Partners", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": false }, { "id": "cmu-berkeley-agent-injection-audit", "title": "Indirect Prompt Injection and Tool Poisoning in Autonomous Multi-Turn Agents: Empirical Vulnerability Assessment and Structural Defenses", "publisher": "Carnegie Mellon University & UC Berkeley", "grade": "A", "score": 93, "freshnessStatus": "current", "critical": false }, { "id": "osworld-webarena-benchmark-audit", "title": "OSWorld and WebArena Autonomous Agent Evaluation Leaderboards and Dynamic Environment Test Logs", "publisher": "OSWorld / WebArena Evaluation Consortium", "grade": "A", "score": 93, "freshnessStatus": "current", "critical": false }, { "id": "eth-berkeley-speculative-leak-paper", "title": "Speculative Leak: Microarchitectural Side-Channels and Cache Contention in Speculative Decoding Inference Engines (CVE-2026-38192)", "publisher": "ETH Zurich & UC Berkeley Systems and Security Lab", "grade": "A", "score": 93, "freshnessStatus": "current", "critical": false }, { "id": "swe-bench-verified-diagnostic-audit", "title": "Diagnostic Failure Modes of Autonomous Coding Agents on SWE-bench Verified: The State Synchronization Bottleneck", "publisher": "Open-Source Software Engineering Evaluation Collective", "grade": "A", "score": 93, "freshnessStatus": "current", "critical": false }, { "id": "bair-stanford-quantization-safety-drift", "title": "Asymmetric Representation Degradation: Why Extreme Quantization Erodes Safety Alignment Faster Than General Intelligence", "publisher": "Berkeley AI Research (BAIR) & Stanford Center for Research on Foundation Models (CRFM)", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": false }, { "id": "google-gemini-2-5-pro-reasoner-paper", "title": "Gemini 2.5 Pro Reasoner Technical Report: Step-Level Supervision, Monte Carlo Tree Search, and Robotics Foundation API", "publisher": "Google DeepMind", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "mistral-large-3-hybrid-ssm-paper", "title": "Mistral Large 3: Dense 145B Hybrid Selective State-Space and Transformer Model Architecture", "publisher": "Mistral AI Research", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "anthropic-multimodal-prompt-smuggling", "title": "Multi-Modal Boundary Breaches in Autonomous Execution Stacks: Exploiting Visual Representation Layers for Agent Privilege Escalation", "publisher": "Anthropic Alignment & Red-Teaming Research", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": false }, { "id": "aisi-joint-multimodal-agent-sandbox-audit", "title": "US and UK AI Safety Institutes: Joint Empirical Audit on Multimodal Visual Prompt Smuggling and Mandatory Sandbox Virtualization for Desktop Agents", "publisher": "US AI Safety Institute (NIST) / UK AI Safety Institute (DSIT)", "grade": "C", "score": 71, "freshnessStatus": "current", "critical": true }, { "id": "meta-llama-4-speculative-decoding-preview", "title": "Meta AI Engineering: Llama 4 Architectural Blueprint, Native Speculative Decoding, and MoE Routing at Scale", "publisher": "Meta AI / Fundamental AI Research", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "nvidia-tensorrt-llm-fp4-blackwell", "title": "NVIDIA TensorRT-LLM 0.16 General Availability: FP4 Microscaling Acceleration on Blackwell GB200 Systems", "publisher": "NVIDIA Technical Blog", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": false }, { "id": "nvidia-blackwell-ultra-nvl72-fp4", "title": "NVIDIA Blackwell Ultra GB200 NVL72 System Architecture: 3.2x Multi-Turn Chain-of-Thought Inference via FP4 Microscaling", "publisher": "NVIDIA Technical Blog", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "deepseek-ascend-cann9-cluster-whitepaper", "title": "DeepSeek 160,000-Accelerator Megacluster on Huawei Ascend CANN 9.0 Architecture", "publisher": "DeepSeek AI / Industry Filings", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true } ], "bestGrade": "A", "worstGrade": "C", "coverage": { "status": "supported", "reason": "At least one valid source explicitly supports this claim." } }, { "id": "claim:benchmark.items.7", "path": "benchmark.items.7", "section": "benchmarks", "kind": "numeric-record", "critical": true, "text": "Gemini 3.1 Pro Preview — score: 46.5", "sourceIds": [ "openai-gpt56", "anthropic-claude-3-7-sonnet-system-card", "openai-o3-pro-system-card", "xai-grok45", "moonshot-kimi-k3", "zhipu-glm52", "alibaba-qwen37", "alibaba-qwen-2-5-coder-reasoner-paper", "swe-bench-verified-october-audit", "google-gemini35", "zhipu-glm-ocr", "baidu-ernie50", "openai-operator-sandbox-architecture", "nist-aisi-agentic-privilege-escalation-bench", "anthropic-claude-3-8-system-card", "deepseek-v3-5-architecture-report", "anthropic-claude-motion-usage-policy", "meta-muse-windows-native", "deepseek-v4", "qwen-audio3", "amd-anthropic", "reuters-aisi-agent-evals-aug5", "arcprize-opus5-jul24", "arcprize-gpt56-jul9", "artificialanalysis-v4flash-aug9", "aisi-test-time-compute", "cvpr-agentsafe-2026", "acl-safeagent-2026", "arxiv-runtime-contract-aug11", "reuters-zai-glm53-aug14", "arxiv-multi-agent-institutions-aug10", "uscc-two-loops-mar23", "reuters-unitree-us-research-aug18", "openai-third-party-cyber-aug4", "jll-emea-dc-aug3", "reuters-aisi-agent-aug20", "arc-anthropic-cot-faithfulness", "llama-3-4-moe-release", "stanford-berkeley-steering", "bird-sql-evaluation-harness", "kv-cache-mla-systems", "mit-cmu-decomposed-semantic-inversion", "swebench-pro-evaluation", "nvidia-rubin-nvl72-infrastructure", "project-glasswing-safeguards", "cmu-berkeley-agent-injection-audit", "osworld-webarena-benchmark-audit", "eth-berkeley-speculative-leak-paper", "swe-bench-verified-diagnostic-audit", "bair-stanford-quantization-safety-drift", "google-gemini-2-5-pro-reasoner-paper", "mistral-large-3-hybrid-ssm-paper", "anthropic-multimodal-prompt-smuggling", "aisi-joint-multimodal-agent-sandbox-audit", "meta-llama-4-speculative-decoding-preview", "nvidia-tensorrt-llm-fp4-blackwell", "nvidia-blackwell-ultra-nvl72-fp4", "deepseek-ascend-cann9-cluster-whitepaper" ], "citationOrigin": "explicit-claim-map", "evidenceMode": "registry-review", "evidenceNote": null, "numericValues": [ "46.5" ], "isNumeric": true, "citationCount": 58, "validSourceIds": [ "openai-gpt56", "anthropic-claude-3-7-sonnet-system-card", "openai-o3-pro-system-card", "xai-grok45", "moonshot-kimi-k3", "zhipu-glm52", "alibaba-qwen37", "alibaba-qwen-2-5-coder-reasoner-paper", "swe-bench-verified-october-audit", "google-gemini35", "zhipu-glm-ocr", "baidu-ernie50", "openai-operator-sandbox-architecture", "nist-aisi-agentic-privilege-escalation-bench", "anthropic-claude-3-8-system-card", "deepseek-v3-5-architecture-report", "anthropic-claude-motion-usage-policy", "meta-muse-windows-native", "deepseek-v4", "qwen-audio3", "amd-anthropic", "reuters-aisi-agent-evals-aug5", "arcprize-opus5-jul24", "arcprize-gpt56-jul9", "artificialanalysis-v4flash-aug9", "aisi-test-time-compute", "cvpr-agentsafe-2026", "acl-safeagent-2026", "arxiv-runtime-contract-aug11", "reuters-zai-glm53-aug14", "arxiv-multi-agent-institutions-aug10", "uscc-two-loops-mar23", "reuters-unitree-us-research-aug18", "openai-third-party-cyber-aug4", "jll-emea-dc-aug3", "reuters-aisi-agent-aug20", "arc-anthropic-cot-faithfulness", "llama-3-4-moe-release", "stanford-berkeley-steering", "bird-sql-evaluation-harness", "kv-cache-mla-systems", "mit-cmu-decomposed-semantic-inversion", "swebench-pro-evaluation", "nvidia-rubin-nvl72-infrastructure", "project-glasswing-safeguards", "cmu-berkeley-agent-injection-audit", "osworld-webarena-benchmark-audit", "eth-berkeley-speculative-leak-paper", "swe-bench-verified-diagnostic-audit", "bair-stanford-quantization-safety-drift", "google-gemini-2-5-pro-reasoner-paper", "mistral-large-3-hybrid-ssm-paper", "anthropic-multimodal-prompt-smuggling", "aisi-joint-multimodal-agent-sandbox-audit", "meta-llama-4-speculative-decoding-preview", "nvidia-tensorrt-llm-fp4-blackwell", "nvidia-blackwell-ultra-nvl72-fp4", "deepseek-ascend-cann9-cluster-whitepaper" ], "invalidSourceIds": [], "sourceDetails": [ { "id": "openai-gpt56", "title": "Introducing GPT-5.6", "publisher": "OpenAI", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true }, { "id": "anthropic-claude-3-7-sonnet-system-card", "title": "Anthropic Claude 3.7 Sonnet: Hybrid Reasoning Architecture, Extended Thinking Token Scaling, and Sub-Pixel Computer Use v2", "publisher": "Anthropic", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "openai-o3-pro-system-card", "title": "OpenAI o3-pro System Card: Frontier Reasoning, Test-Time Tree Search with Native Multi-Step Tool Execution, and Autonomous Coding Evaluations", "publisher": "OpenAI", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "xai-grok45", "title": "Grok 4.5", "publisher": "xAI", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true }, { "id": "moonshot-kimi-k3", "title": "Kimi API platform — Kimi K3 model and pricing", "publisher": "Moonshot AI", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true }, { "id": "zhipu-glm52", "title": "Zhipu BigModel — GLM-5.2 model overview", "publisher": "Zhipu AI", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true }, { "id": "alibaba-qwen37", "title": "Alibaba Cloud Model Studio — Qwen flagship models", "publisher": "Alibaba Cloud", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true }, { "id": "alibaba-qwen-2-5-coder-reasoner-paper", "title": "Qwen 2.5-Coder-32B-Reasoner: Open-Weight Test-Time Search Code Agent with Step-Level Process Reward Supervision", "publisher": "Alibaba Cloud & Qwen Team", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "swe-bench-verified-october-audit", "title": "SWE-bench Verified & Terminal-Bench October 2026 Frontier Coding Leaderboard: Test-Time Reasoning vs. Open-Weight Architectures", "publisher": "CodeEval Foundation / SWE-bench Consortium", "grade": "A", "score": 97, "freshnessStatus": "current", "critical": true }, { "id": "google-gemini35", "title": "Gemini 3.5", "publisher": "Google", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true }, { "id": "zhipu-glm-ocr", "title": "Zhipu BigModel — GLM-OCR", "publisher": "Zhipu AI", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "baidu-ernie50", "title": "Baidu Qianfan — ERNIE 5.0 model list and international price", "publisher": "Baidu AI Cloud", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true }, { "id": "openai-operator-sandbox-architecture", "title": "OpenAI Operator Architecture: MicroVM Hardware Isolation, Permission Gating, and Dual-Loop Verification for Web Agents", "publisher": "OpenAI", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "nist-aisi-agentic-privilege-escalation-bench", "title": "US Artificial Intelligence Safety Institute (NIST AISI): Standardized Test Suite for Agentic Privilege Escalation and MicroVM Containment", "publisher": "National Institute of Standards and Technology (NIST)", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "anthropic-claude-3-8-system-card", "title": "Anthropic Claude 3.8 Architecture & System Card: Computer Use v2.5, POSIX Bash Sandboxing, and SWE-bench Verified Benchmark", "publisher": "Anthropic PBC", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "deepseek-v3-5-architecture-report", "title": "DeepSeek-V3.5 Technical Report: Multi-Head Latent Attention (MLA) Scaling and DualPipe Parallelism across 2,048 Compute Nodes", "publisher": "DeepSeek AI", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "anthropic-claude-motion-usage-policy", "title": "Anthropic Claude Motion Engine, Dashboards Beta, and System Call Containment Policy", "publisher": "Anthropic PBC", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "meta-muse-windows-native", "title": "Meta Muse Desktop Architecture: Native Windows Runtime, Local Memory Buffers, and Cross-Platform Integration", "publisher": "Meta Platforms & Microsoft", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "deepseek-v4", "title": "DeepSeek V4 Preview", "publisher": "DeepSeek", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true }, { "id": "qwen-audio3", "title": "Qwen-Audio-3.0-TTS: More Multilingual, Easier to Direct", "publisher": "Alibaba Cloud / Qwen", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "amd-anthropic", "title": "AMD and Anthropic announce strategic partnership for up to 2 GW of MI450 Series GPUs", "publisher": "AMD Newsroom", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "reuters-aisi-agent-evals-aug5", "title": "OpenAI, Anthropic AI agents implicated in new security breaches", "publisher": "Reuters", "grade": "C", "score": 71, "freshnessStatus": "current", "critical": false }, { "id": "arcprize-opus5-jul24", "title": "Claude Opus 5 — ARC-AGI results", "publisher": "ARC Prize", "grade": "A", "score": 93, "freshnessStatus": "current", "critical": false }, { "id": "arcprize-gpt56-jul9", "title": "GPT-5.6 — ARC-AGI results", "publisher": "ARC Prize", "grade": "A", "score": 93, "freshnessStatus": "current", "critical": false }, { "id": "artificialanalysis-v4flash-aug9", "title": "DeepSeek V4 Flash — independent model analysis", "publisher": "Artificial Analysis", "grade": "C", "score": 71, "freshnessStatus": "current", "critical": false }, { "id": "aisi-test-time-compute", "title": "Measuring AI agents progress on multi-step cyber attack scenarios", "publisher": "UK AI Security Institute", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": false }, { "id": "cvpr-agentsafe-2026", "title": "AGENTSAFE: Benchmarking the Safety of Embodied Agents on Hazardous Instructions", "publisher": "CVPR 2026", "grade": "B", "score": 85, "freshnessStatus": "current", "critical": false }, { "id": "acl-safeagent-2026", "title": "SafeAgent: Safeguarding LLM Agents via an Automated Risk Simulator", "publisher": "ACL 2026", "grade": "B", "score": 85, "freshnessStatus": "current", "critical": false }, { "id": "arxiv-runtime-contract-aug11", "title": "Agent Safety Should Be a Runtime Contract", "publisher": "arXiv", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": false }, { "id": "reuters-zai-glm53-aug14", "title": "China's Z.ai says new model nears Anthropic's Mythos 5 in cyber-defence tests", "publisher": "Reuters", "grade": "A", "score": 93, "freshnessStatus": "current", "critical": false }, { "id": "arxiv-multi-agent-institutions-aug10", "title": "Multi-Agent AI Safety as an Institutional Design Problem", "publisher": "arXiv", "grade": "A", "score": 93, "freshnessStatus": "current", "critical": false }, { "id": "uscc-two-loops-mar23", "title": "Two Loops: How China’s Open AI Strategy Reinforces Its Industrial Dominance", "publisher": "U.S.-China Economic and Security Review Commission", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": false }, { "id": "reuters-unitree-us-research-aug18", "title": "How US military funding propelled China’s robot dogs", "publisher": "Reuters", "grade": "A", "score": 93, "freshnessStatus": "current", "critical": false }, { "id": "openai-third-party-cyber-aug4", "title": "Third-party cyber evaluations involving OpenAI models", "publisher": "OpenAI", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": false }, { "id": "jll-emea-dc-aug3", "title": "EMEA data centre mid-year 2026 report", "publisher": "JLL", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": false }, { "id": "reuters-aisi-agent-aug20", "title": "How a Texas student blew the whistle on a rogue AI hacking attempt", "publisher": "Reuters", "grade": "C", "score": 71, "freshnessStatus": "current", "critical": false }, { "id": "arc-anthropic-cot-faithfulness", "title": "Evaluating Chain-of-Thought Faithfulness Under Outcome Reinforcement Learning", "publisher": "Anthropic Research & Alignment Research Center", "grade": "A", "score": 93, "freshnessStatus": "current", "critical": false }, { "id": "llama-3-4-moe-release", "title": "Meta Llama 3.4-MoE Weights Release & Speculative Draft Heads", "publisher": "Meta AI / Hugging Face", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": false }, { "id": "stanford-berkeley-steering", "title": "Deterministic Containment of Autonomous Agent Tool Invocation via Residual Activation Steering", "publisher": "Stanford CRFM & UC Berkeley AI Research", "grade": "A", "score": 93, "freshnessStatus": "current", "critical": false }, { "id": "bird-sql-evaluation-harness", "title": "BIRD-SQL Cross-Domain Enterprise Relational Database Benchmark Leaderboard", "publisher": "BIRD-SQL Consortium / DAMO", "grade": "A", "score": 93, "freshnessStatus": "current", "critical": false }, { "id": "kv-cache-mla-systems", "title": "Multi-Head Latent Attention and 4-Bit PagedAttention Serving Implementations", "publisher": "vLLM Project & TensorRT-LLM", "grade": "A", "score": 93, "freshnessStatus": "current", "critical": false }, { "id": "mit-cmu-decomposed-semantic-inversion", "title": "Decomposed Semantic Inversion in Long-Context LLMs", "publisher": "MIT CSAIL & Carnegie Mellon University", "grade": "A", "score": 93, "freshnessStatus": "current", "critical": false }, { "id": "swebench-pro-evaluation", "title": "SWE-bench Verified Saturation Analysis and SWE-bench Pro Evaluation Methodology", "publisher": "SWE-bench Consortium", "grade": "A", "score": 93, "freshnessStatus": "current", "critical": false }, { "id": "nvidia-rubin-nvl72-infrastructure", "title": "NVIDIA Rubin NVL72 and Blackwell Ultra Rack-Scale Architecture Specifications", "publisher": "NVIDIA Corporation", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": false }, { "id": "project-glasswing-safeguards", "title": "Project Glasswing Defensive Cybersecurity Verification Framework and Enterprise Frontier Safeguards", "publisher": "Anthropic & Coalition Partners", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": false }, { "id": "cmu-berkeley-agent-injection-audit", "title": "Indirect Prompt Injection and Tool Poisoning in Autonomous Multi-Turn Agents: Empirical Vulnerability Assessment and Structural Defenses", "publisher": "Carnegie Mellon University & UC Berkeley", "grade": "A", "score": 93, "freshnessStatus": "current", "critical": false }, { "id": "osworld-webarena-benchmark-audit", "title": "OSWorld and WebArena Autonomous Agent Evaluation Leaderboards and Dynamic Environment Test Logs", "publisher": "OSWorld / WebArena Evaluation Consortium", "grade": "A", "score": 93, "freshnessStatus": "current", "critical": false }, { "id": "eth-berkeley-speculative-leak-paper", "title": "Speculative Leak: Microarchitectural Side-Channels and Cache Contention in Speculative Decoding Inference Engines (CVE-2026-38192)", "publisher": "ETH Zurich & UC Berkeley Systems and Security Lab", "grade": "A", "score": 93, "freshnessStatus": "current", "critical": false }, { "id": "swe-bench-verified-diagnostic-audit", "title": "Diagnostic Failure Modes of Autonomous Coding Agents on SWE-bench Verified: The State Synchronization Bottleneck", "publisher": "Open-Source Software Engineering Evaluation Collective", "grade": "A", "score": 93, "freshnessStatus": "current", "critical": false }, { "id": "bair-stanford-quantization-safety-drift", "title": "Asymmetric Representation Degradation: Why Extreme Quantization Erodes Safety Alignment Faster Than General Intelligence", "publisher": "Berkeley AI Research (BAIR) & Stanford Center for Research on Foundation Models (CRFM)", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": false }, { "id": "google-gemini-2-5-pro-reasoner-paper", "title": "Gemini 2.5 Pro Reasoner Technical Report: Step-Level Supervision, Monte Carlo Tree Search, and Robotics Foundation API", "publisher": "Google DeepMind", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "mistral-large-3-hybrid-ssm-paper", "title": "Mistral Large 3: Dense 145B Hybrid Selective State-Space and Transformer Model Architecture", "publisher": "Mistral AI Research", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "anthropic-multimodal-prompt-smuggling", "title": "Multi-Modal Boundary Breaches in Autonomous Execution Stacks: Exploiting Visual Representation Layers for Agent Privilege Escalation", "publisher": "Anthropic Alignment & Red-Teaming Research", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": false }, { "id": "aisi-joint-multimodal-agent-sandbox-audit", "title": "US and UK AI Safety Institutes: Joint Empirical Audit on Multimodal Visual Prompt Smuggling and Mandatory Sandbox Virtualization for Desktop Agents", "publisher": "US AI Safety Institute (NIST) / UK AI Safety Institute (DSIT)", "grade": "C", "score": 71, "freshnessStatus": "current", "critical": true }, { "id": "meta-llama-4-speculative-decoding-preview", "title": "Meta AI Engineering: Llama 4 Architectural Blueprint, Native Speculative Decoding, and MoE Routing at Scale", "publisher": "Meta AI / Fundamental AI Research", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "nvidia-tensorrt-llm-fp4-blackwell", "title": "NVIDIA TensorRT-LLM 0.16 General Availability: FP4 Microscaling Acceleration on Blackwell GB200 Systems", "publisher": "NVIDIA Technical Blog", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": false }, { "id": "nvidia-blackwell-ultra-nvl72-fp4", "title": "NVIDIA Blackwell Ultra GB200 NVL72 System Architecture: 3.2x Multi-Turn Chain-of-Thought Inference via FP4 Microscaling", "publisher": "NVIDIA Technical Blog", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "deepseek-ascend-cann9-cluster-whitepaper", "title": "DeepSeek 160,000-Accelerator Megacluster on Huawei Ascend CANN 9.0 Architecture", "publisher": "DeepSeek AI / Industry Filings", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true } ], "bestGrade": "A", "worstGrade": "C", "coverage": { "status": "supported", "reason": "At least one valid source explicitly supports this claim." } }, { "id": "claim:benchmarkDimensions.0.status", "path": "benchmarkDimensions.0.status", "section": "benchmarks", "kind": "verification-claim", "critical": true, "text": "General intelligence and reasoning: Verified snapshot available; no new comparable July 22 evaluation verified", "sourceIds": [ "openai-gpt56", "anthropic-claude-3-7-sonnet-system-card", "openai-o3-pro-system-card" ], "citationOrigin": "explicit-claim-map", "evidenceMode": "registry-review", "evidenceNote": null, "numericValues": [], "isNumeric": false, "citationCount": 3, "validSourceIds": [ "openai-gpt56", "anthropic-claude-3-7-sonnet-system-card", "openai-o3-pro-system-card" ], "invalidSourceIds": [], "sourceDetails": [ { "id": "openai-gpt56", "title": "Introducing GPT-5.6", "publisher": "OpenAI", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true }, { "id": "anthropic-claude-3-7-sonnet-system-card", "title": "Anthropic Claude 3.7 Sonnet: Hybrid Reasoning Architecture, Extended Thinking Token Scaling, and Sub-Pixel Computer Use v2", "publisher": "Anthropic", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "openai-o3-pro-system-card", "title": "OpenAI o3-pro System Card: Frontier Reasoning, Test-Time Tree Search with Native Multi-Step Tool Execution, and Autonomous Coding Evaluations", "publisher": "OpenAI", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true } ], "bestGrade": "A", "worstGrade": "B", "coverage": { "status": "supported", "reason": "At least one valid source explicitly supports this claim." } }, { "id": "claim:benchmarkDimensions.0.items.0", "path": "benchmarkDimensions.0.items.0", "section": "benchmarks", "kind": "numeric-record", "critical": true, "text": "Claude Fable 5 — score: 59.9", "sourceIds": [ "openai-gpt56", "anthropic-claude-3-7-sonnet-system-card", "openai-o3-pro-system-card" ], "citationOrigin": "explicit-claim-map", "evidenceMode": "registry-review", "evidenceNote": null, "numericValues": [ "59.9" ], "isNumeric": true, "citationCount": 3, "validSourceIds": [ "openai-gpt56", "anthropic-claude-3-7-sonnet-system-card", "openai-o3-pro-system-card" ], "invalidSourceIds": [], "sourceDetails": [ { "id": "openai-gpt56", "title": "Introducing GPT-5.6", "publisher": "OpenAI", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true }, { "id": "anthropic-claude-3-7-sonnet-system-card", "title": "Anthropic Claude 3.7 Sonnet: Hybrid Reasoning Architecture, Extended Thinking Token Scaling, and Sub-Pixel Computer Use v2", "publisher": "Anthropic", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "openai-o3-pro-system-card", "title": "OpenAI o3-pro System Card: Frontier Reasoning, Test-Time Tree Search with Native Multi-Step Tool Execution, and Autonomous Coding Evaluations", "publisher": "OpenAI", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true } ], "bestGrade": "A", "worstGrade": "B", "coverage": { "status": "supported", "reason": "At least one valid source explicitly supports this claim." } }, { "id": "claim:benchmarkDimensions.0.items.1", "path": "benchmarkDimensions.0.items.1", "section": "benchmarks", "kind": "numeric-record", "critical": true, "text": "GPT-5.6 Sol — score: 58.9", "sourceIds": [ "openai-gpt56", "anthropic-claude-3-7-sonnet-system-card", "openai-o3-pro-system-card" ], "citationOrigin": "explicit-claim-map", "evidenceMode": "registry-review", "evidenceNote": null, "numericValues": [ "58.9" ], "isNumeric": true, "citationCount": 3, "validSourceIds": [ "openai-gpt56", "anthropic-claude-3-7-sonnet-system-card", "openai-o3-pro-system-card" ], "invalidSourceIds": [], "sourceDetails": [ { "id": "openai-gpt56", "title": "Introducing GPT-5.6", "publisher": "OpenAI", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true }, { "id": "anthropic-claude-3-7-sonnet-system-card", "title": "Anthropic Claude 3.7 Sonnet: Hybrid Reasoning Architecture, Extended Thinking Token Scaling, and Sub-Pixel Computer Use v2", "publisher": "Anthropic", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "openai-o3-pro-system-card", "title": "OpenAI o3-pro System Card: Frontier Reasoning, Test-Time Tree Search with Native Multi-Step Tool Execution, and Autonomous Coding Evaluations", "publisher": "OpenAI", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true } ], "bestGrade": "A", "worstGrade": "B", "coverage": { "status": "supported", "reason": "At least one valid source explicitly supports this claim." } }, { "id": "claim:benchmarkDimensions.0.items.2", "path": "benchmarkDimensions.0.items.2", "section": "benchmarks", "kind": "numeric-record", "critical": true, "text": "Claude Opus 4.8 — score: 55.7", "sourceIds": [ "openai-gpt56", "anthropic-claude-3-7-sonnet-system-card", "openai-o3-pro-system-card" ], "citationOrigin": "explicit-claim-map", "evidenceMode": "registry-review", "evidenceNote": null, "numericValues": [ "55.7" ], "isNumeric": true, "citationCount": 3, "validSourceIds": [ "openai-gpt56", "anthropic-claude-3-7-sonnet-system-card", "openai-o3-pro-system-card" ], "invalidSourceIds": [], "sourceDetails": [ { "id": "openai-gpt56", "title": "Introducing GPT-5.6", "publisher": "OpenAI", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true }, { "id": "anthropic-claude-3-7-sonnet-system-card", "title": "Anthropic Claude 3.7 Sonnet: Hybrid Reasoning Architecture, Extended Thinking Token Scaling, and Sub-Pixel Computer Use v2", "publisher": "Anthropic", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "openai-o3-pro-system-card", "title": "OpenAI o3-pro System Card: Frontier Reasoning, Test-Time Tree Search with Native Multi-Step Tool Execution, and Autonomous Coding Evaluations", "publisher": "OpenAI", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true } ], "bestGrade": "A", "worstGrade": "B", "coverage": { "status": "supported", "reason": "At least one valid source explicitly supports this claim." } }, { "id": "claim:benchmarkDimensions.0.items.3", "path": "benchmarkDimensions.0.items.3", "section": "benchmarks", "kind": "numeric-record", "critical": true, "text": "GPT-5.5 — score: 54.8", "sourceIds": [ "openai-gpt56", "anthropic-claude-3-7-sonnet-system-card", "openai-o3-pro-system-card" ], "citationOrigin": "explicit-claim-map", "evidenceMode": "registry-review", "evidenceNote": null, "numericValues": [ "54.8" ], "isNumeric": true, "citationCount": 3, "validSourceIds": [ "openai-gpt56", "anthropic-claude-3-7-sonnet-system-card", "openai-o3-pro-system-card" ], "invalidSourceIds": [], "sourceDetails": [ { "id": "openai-gpt56", "title": "Introducing GPT-5.6", "publisher": "OpenAI", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true }, { "id": "anthropic-claude-3-7-sonnet-system-card", "title": "Anthropic Claude 3.7 Sonnet: Hybrid Reasoning Architecture, Extended Thinking Token Scaling, and Sub-Pixel Computer Use v2", "publisher": "Anthropic", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "openai-o3-pro-system-card", "title": "OpenAI o3-pro System Card: Frontier Reasoning, Test-Time Tree Search with Native Multi-Step Tool Execution, and Autonomous Coding Evaluations", "publisher": "OpenAI", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true } ], "bestGrade": "A", "worstGrade": "B", "coverage": { "status": "supported", "reason": "At least one valid source explicitly supports this claim." } }, { "id": "claim:benchmarkDimensions.0.items.4", "path": "benchmarkDimensions.0.items.4", "section": "benchmarks", "kind": "numeric-record", "critical": true, "text": "GPT-5.6 Terra — score: 55", "sourceIds": [ "openai-gpt56", "anthropic-claude-3-7-sonnet-system-card", "openai-o3-pro-system-card" ], "citationOrigin": "explicit-claim-map", "evidenceMode": "registry-review", "evidenceNote": null, "numericValues": [ "55" ], "isNumeric": true, "citationCount": 3, "validSourceIds": [ "openai-gpt56", "anthropic-claude-3-7-sonnet-system-card", "openai-o3-pro-system-card" ], "invalidSourceIds": [], "sourceDetails": [ { "id": "openai-gpt56", "title": "Introducing GPT-5.6", "publisher": "OpenAI", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true }, { "id": "anthropic-claude-3-7-sonnet-system-card", "title": "Anthropic Claude 3.7 Sonnet: Hybrid Reasoning Architecture, Extended Thinking Token Scaling, and Sub-Pixel Computer Use v2", "publisher": "Anthropic", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "openai-o3-pro-system-card", "title": "OpenAI o3-pro System Card: Frontier Reasoning, Test-Time Tree Search with Native Multi-Step Tool Execution, and Autonomous Coding Evaluations", "publisher": "OpenAI", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true } ], "bestGrade": "A", "worstGrade": "B", "coverage": { "status": "supported", "reason": "At least one valid source explicitly supports this claim." } }, { "id": "claim:benchmarkDimensions.0.items.5", "path": "benchmarkDimensions.0.items.5", "section": "benchmarks", "kind": "numeric-record", "critical": true, "text": "GPT-5.6 Luna — score: 51.2", "sourceIds": [ "openai-gpt56", "anthropic-claude-3-7-sonnet-system-card", "openai-o3-pro-system-card" ], "citationOrigin": "explicit-claim-map", "evidenceMode": "registry-review", "evidenceNote": null, "numericValues": [ "51.2" ], "isNumeric": true, "citationCount": 3, "validSourceIds": [ "openai-gpt56", "anthropic-claude-3-7-sonnet-system-card", "openai-o3-pro-system-card" ], "invalidSourceIds": [], "sourceDetails": [ { "id": "openai-gpt56", "title": "Introducing GPT-5.6", "publisher": "OpenAI", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true }, { "id": "anthropic-claude-3-7-sonnet-system-card", "title": "Anthropic Claude 3.7 Sonnet: Hybrid Reasoning Architecture, Extended Thinking Token Scaling, and Sub-Pixel Computer Use v2", "publisher": "Anthropic", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "openai-o3-pro-system-card", "title": "OpenAI o3-pro System Card: Frontier Reasoning, Test-Time Tree Search with Native Multi-Step Tool Execution, and Autonomous Coding Evaluations", "publisher": "OpenAI", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true } ], "bestGrade": "A", "worstGrade": "B", "coverage": { "status": "supported", "reason": "At least one valid source explicitly supports this claim." } }, { "id": "claim:benchmarkDimensions.0.items.6", "path": "benchmarkDimensions.0.items.6", "section": "benchmarks", "kind": "numeric-record", "critical": true, "text": "Gemini 3.5 Flash — score: 50.2", "sourceIds": [ "openai-gpt56", "anthropic-claude-3-7-sonnet-system-card", "openai-o3-pro-system-card" ], "citationOrigin": "explicit-claim-map", "evidenceMode": "registry-review", "evidenceNote": null, "numericValues": [ "50.2" ], "isNumeric": true, "citationCount": 3, "validSourceIds": [ "openai-gpt56", "anthropic-claude-3-7-sonnet-system-card", "openai-o3-pro-system-card" ], "invalidSourceIds": [], "sourceDetails": [ { "id": "openai-gpt56", "title": "Introducing GPT-5.6", "publisher": "OpenAI", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true }, { "id": "anthropic-claude-3-7-sonnet-system-card", "title": "Anthropic Claude 3.7 Sonnet: Hybrid Reasoning Architecture, Extended Thinking Token Scaling, and Sub-Pixel Computer Use v2", "publisher": "Anthropic", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "openai-o3-pro-system-card", "title": "OpenAI o3-pro System Card: Frontier Reasoning, Test-Time Tree Search with Native Multi-Step Tool Execution, and Autonomous Coding Evaluations", "publisher": "OpenAI", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true } ], "bestGrade": "A", "worstGrade": "B", "coverage": { "status": "supported", "reason": "At least one valid source explicitly supports this claim." } }, { "id": "claim:benchmarkDimensions.0.items.7", "path": "benchmarkDimensions.0.items.7", "section": "benchmarks", "kind": "numeric-record", "critical": true, "text": "Gemini 3.1 Pro Preview — score: 46.5", "sourceIds": [ "openai-gpt56", "anthropic-claude-3-7-sonnet-system-card", "openai-o3-pro-system-card" ], "citationOrigin": "explicit-claim-map", "evidenceMode": "registry-review", "evidenceNote": null, "numericValues": [ "46.5" ], "isNumeric": true, "citationCount": 3, "validSourceIds": [ "openai-gpt56", "anthropic-claude-3-7-sonnet-system-card", "openai-o3-pro-system-card" ], "invalidSourceIds": [], "sourceDetails": [ { "id": "openai-gpt56", "title": "Introducing GPT-5.6", "publisher": "OpenAI", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true }, { "id": "anthropic-claude-3-7-sonnet-system-card", "title": "Anthropic Claude 3.7 Sonnet: Hybrid Reasoning Architecture, Extended Thinking Token Scaling, and Sub-Pixel Computer Use v2", "publisher": "Anthropic", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "openai-o3-pro-system-card", "title": "OpenAI o3-pro System Card: Frontier Reasoning, Test-Time Tree Search with Native Multi-Step Tool Execution, and Autonomous Coding Evaluations", "publisher": "OpenAI", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true } ], "bestGrade": "A", "worstGrade": "B", "coverage": { "status": "supported", "reason": "At least one valid source explicitly supports this claim." } }, { "id": "claim:benchmarkDimensions.1.status", "path": "benchmarkDimensions.1.status", "section": "benchmarks", "kind": "verification-claim", "critical": true, "text": "Coding and software engineering: Provider-published comparable scorecards available; no new comparable July 22 evaluation verified", "sourceIds": [ "openai-gpt56", "xai-grok45", "moonshot-kimi-k3", "zhipu-glm52", "alibaba-qwen37", "anthropic-claude-3-7-sonnet-system-card", "openai-o3-pro-system-card", "alibaba-qwen-2-5-coder-reasoner-paper", "swe-bench-verified-october-audit" ], "citationOrigin": "explicit-claim-map", "evidenceMode": "registry-review", "evidenceNote": null, "numericValues": [], "isNumeric": false, "citationCount": 9, "validSourceIds": [ "openai-gpt56", "xai-grok45", "moonshot-kimi-k3", "zhipu-glm52", "alibaba-qwen37", "anthropic-claude-3-7-sonnet-system-card", "openai-o3-pro-system-card", "alibaba-qwen-2-5-coder-reasoner-paper", "swe-bench-verified-october-audit" ], "invalidSourceIds": [], "sourceDetails": [ { "id": "openai-gpt56", "title": "Introducing GPT-5.6", "publisher": "OpenAI", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true }, { "id": "xai-grok45", "title": "Grok 4.5", "publisher": "xAI", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true }, { "id": "moonshot-kimi-k3", "title": "Kimi API platform — Kimi K3 model and pricing", "publisher": "Moonshot AI", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true }, { "id": "zhipu-glm52", "title": "Zhipu BigModel — GLM-5.2 model overview", "publisher": "Zhipu AI", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true }, { "id": "alibaba-qwen37", "title": "Alibaba Cloud Model Studio — Qwen flagship models", "publisher": "Alibaba Cloud", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true }, { "id": "anthropic-claude-3-7-sonnet-system-card", "title": "Anthropic Claude 3.7 Sonnet: Hybrid Reasoning Architecture, Extended Thinking Token Scaling, and Sub-Pixel Computer Use v2", "publisher": "Anthropic", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "openai-o3-pro-system-card", "title": "OpenAI o3-pro System Card: Frontier Reasoning, Test-Time Tree Search with Native Multi-Step Tool Execution, and Autonomous Coding Evaluations", "publisher": "OpenAI", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "alibaba-qwen-2-5-coder-reasoner-paper", "title": "Qwen 2.5-Coder-32B-Reasoner: Open-Weight Test-Time Search Code Agent with Step-Level Process Reward Supervision", "publisher": "Alibaba Cloud & Qwen Team", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "swe-bench-verified-october-audit", "title": "SWE-bench Verified & Terminal-Bench October 2026 Frontier Coding Leaderboard: Test-Time Reasoning vs. Open-Weight Architectures", "publisher": "CodeEval Foundation / SWE-bench Consortium", "grade": "A", "score": 97, "freshnessStatus": "current", "critical": true } ], "bestGrade": "A", "worstGrade": "B", "coverage": { "status": "supported", "reason": "At least one valid source explicitly supports this claim." } }, { "id": "claim:benchmarkDimensions.1.items.0", "path": "benchmarkDimensions.1.items.0", "section": "benchmarks", "kind": "numeric-record", "critical": true, "text": "GPT-5.6 Sol — score: 88.8", "sourceIds": [ "openai-gpt56", "xai-grok45", "moonshot-kimi-k3", "zhipu-glm52", "alibaba-qwen37", "anthropic-claude-3-7-sonnet-system-card", "openai-o3-pro-system-card", "alibaba-qwen-2-5-coder-reasoner-paper", "swe-bench-verified-october-audit" ], "citationOrigin": "explicit-claim-map", "evidenceMode": "registry-review", "evidenceNote": null, "numericValues": [ "88.8" ], "isNumeric": true, "citationCount": 9, "validSourceIds": [ "openai-gpt56", "xai-grok45", "moonshot-kimi-k3", "zhipu-glm52", "alibaba-qwen37", "anthropic-claude-3-7-sonnet-system-card", "openai-o3-pro-system-card", "alibaba-qwen-2-5-coder-reasoner-paper", "swe-bench-verified-october-audit" ], "invalidSourceIds": [], "sourceDetails": [ { "id": "openai-gpt56", "title": "Introducing GPT-5.6", "publisher": "OpenAI", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true }, { "id": "xai-grok45", "title": "Grok 4.5", "publisher": "xAI", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true }, { "id": "moonshot-kimi-k3", "title": "Kimi API platform — Kimi K3 model and pricing", "publisher": "Moonshot AI", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true }, { "id": "zhipu-glm52", "title": "Zhipu BigModel — GLM-5.2 model overview", "publisher": "Zhipu AI", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true }, { "id": "alibaba-qwen37", "title": "Alibaba Cloud Model Studio — Qwen flagship models", "publisher": "Alibaba Cloud", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true }, { "id": "anthropic-claude-3-7-sonnet-system-card", "title": "Anthropic Claude 3.7 Sonnet: Hybrid Reasoning Architecture, Extended Thinking Token Scaling, and Sub-Pixel Computer Use v2", "publisher": "Anthropic", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "openai-o3-pro-system-card", "title": "OpenAI o3-pro System Card: Frontier Reasoning, Test-Time Tree Search with Native Multi-Step Tool Execution, and Autonomous Coding Evaluations", "publisher": "OpenAI", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "alibaba-qwen-2-5-coder-reasoner-paper", "title": "Qwen 2.5-Coder-32B-Reasoner: Open-Weight Test-Time Search Code Agent with Step-Level Process Reward Supervision", "publisher": "Alibaba Cloud & Qwen Team", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "swe-bench-verified-october-audit", "title": "SWE-bench Verified & Terminal-Bench October 2026 Frontier Coding Leaderboard: Test-Time Reasoning vs. Open-Weight Architectures", "publisher": "CodeEval Foundation / SWE-bench Consortium", "grade": "A", "score": 97, "freshnessStatus": "current", "critical": true } ], "bestGrade": "A", "worstGrade": "B", "coverage": { "status": "supported", "reason": "At least one valid source explicitly supports this claim." } }, { "id": "claim:benchmarkDimensions.1.items.1", "path": "benchmarkDimensions.1.items.1", "section": "benchmarks", "kind": "numeric-record", "critical": true, "text": "Claude Mythos 5 — score: 88", "sourceIds": [ "openai-gpt56", "xai-grok45", "moonshot-kimi-k3", "zhipu-glm52", "alibaba-qwen37", "anthropic-claude-3-7-sonnet-system-card", "openai-o3-pro-system-card", "alibaba-qwen-2-5-coder-reasoner-paper", "swe-bench-verified-october-audit" ], "citationOrigin": "explicit-claim-map", "evidenceMode": "registry-review", "evidenceNote": null, "numericValues": [ "88" ], "isNumeric": true, "citationCount": 9, "validSourceIds": [ "openai-gpt56", "xai-grok45", "moonshot-kimi-k3", "zhipu-glm52", "alibaba-qwen37", "anthropic-claude-3-7-sonnet-system-card", "openai-o3-pro-system-card", "alibaba-qwen-2-5-coder-reasoner-paper", "swe-bench-verified-october-audit" ], "invalidSourceIds": [], "sourceDetails": [ { "id": "openai-gpt56", "title": "Introducing GPT-5.6", "publisher": "OpenAI", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true }, { "id": "xai-grok45", "title": "Grok 4.5", "publisher": "xAI", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true }, { "id": "moonshot-kimi-k3", "title": "Kimi API platform — Kimi K3 model and pricing", "publisher": "Moonshot AI", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true }, { "id": "zhipu-glm52", "title": "Zhipu BigModel — GLM-5.2 model overview", "publisher": "Zhipu AI", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true }, { "id": "alibaba-qwen37", "title": "Alibaba Cloud Model Studio — Qwen flagship models", "publisher": "Alibaba Cloud", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true }, { "id": "anthropic-claude-3-7-sonnet-system-card", "title": "Anthropic Claude 3.7 Sonnet: Hybrid Reasoning Architecture, Extended Thinking Token Scaling, and Sub-Pixel Computer Use v2", "publisher": "Anthropic", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "openai-o3-pro-system-card", "title": "OpenAI o3-pro System Card: Frontier Reasoning, Test-Time Tree Search with Native Multi-Step Tool Execution, and Autonomous Coding Evaluations", "publisher": "OpenAI", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "alibaba-qwen-2-5-coder-reasoner-paper", "title": "Qwen 2.5-Coder-32B-Reasoner: Open-Weight Test-Time Search Code Agent with Step-Level Process Reward Supervision", "publisher": "Alibaba Cloud & Qwen Team", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "swe-bench-verified-october-audit", "title": "SWE-bench Verified & Terminal-Bench October 2026 Frontier Coding Leaderboard: Test-Time Reasoning vs. Open-Weight Architectures", "publisher": "CodeEval Foundation / SWE-bench Consortium", "grade": "A", "score": 97, "freshnessStatus": "current", "critical": true } ], "bestGrade": "A", "worstGrade": "B", "coverage": { "status": "supported", "reason": "At least one valid source explicitly supports this claim." } }, { "id": "claim:benchmarkDimensions.1.items.2", "path": "benchmarkDimensions.1.items.2", "section": "benchmarks", "kind": "numeric-record", "critical": true, "text": "GPT-5.6 Terra — score: 87.4", "sourceIds": [ "openai-gpt56", "xai-grok45", "moonshot-kimi-k3", "zhipu-glm52", "alibaba-qwen37", "anthropic-claude-3-7-sonnet-system-card", "openai-o3-pro-system-card", "alibaba-qwen-2-5-coder-reasoner-paper", "swe-bench-verified-october-audit" ], "citationOrigin": "explicit-claim-map", "evidenceMode": "registry-review", "evidenceNote": null, "numericValues": [ "87.4" ], "isNumeric": true, "citationCount": 9, "validSourceIds": [ "openai-gpt56", "xai-grok45", "moonshot-kimi-k3", "zhipu-glm52", "alibaba-qwen37", "anthropic-claude-3-7-sonnet-system-card", "openai-o3-pro-system-card", "alibaba-qwen-2-5-coder-reasoner-paper", "swe-bench-verified-october-audit" ], "invalidSourceIds": [], "sourceDetails": [ { "id": "openai-gpt56", "title": "Introducing GPT-5.6", "publisher": "OpenAI", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true }, { "id": "xai-grok45", "title": "Grok 4.5", "publisher": "xAI", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true }, { "id": "moonshot-kimi-k3", "title": "Kimi API platform — Kimi K3 model and pricing", "publisher": "Moonshot AI", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true }, { "id": "zhipu-glm52", "title": "Zhipu BigModel — GLM-5.2 model overview", "publisher": "Zhipu AI", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true }, { "id": "alibaba-qwen37", "title": "Alibaba Cloud Model Studio — Qwen flagship models", "publisher": "Alibaba Cloud", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true }, { "id": "anthropic-claude-3-7-sonnet-system-card", "title": "Anthropic Claude 3.7 Sonnet: Hybrid Reasoning Architecture, Extended Thinking Token Scaling, and Sub-Pixel Computer Use v2", "publisher": "Anthropic", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "openai-o3-pro-system-card", "title": "OpenAI o3-pro System Card: Frontier Reasoning, Test-Time Tree Search with Native Multi-Step Tool Execution, and Autonomous Coding Evaluations", "publisher": "OpenAI", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "alibaba-qwen-2-5-coder-reasoner-paper", "title": "Qwen 2.5-Coder-32B-Reasoner: Open-Weight Test-Time Search Code Agent with Step-Level Process Reward Supervision", "publisher": "Alibaba Cloud & Qwen Team", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "swe-bench-verified-october-audit", "title": "SWE-bench Verified & Terminal-Bench October 2026 Frontier Coding Leaderboard: Test-Time Reasoning vs. Open-Weight Architectures", "publisher": "CodeEval Foundation / SWE-bench Consortium", "grade": "A", "score": 97, "freshnessStatus": "current", "critical": true } ], "bestGrade": "A", "worstGrade": "B", "coverage": { "status": "supported", "reason": "At least one valid source explicitly supports this claim." } }, { "id": "claim:benchmarkDimensions.1.items.3", "path": "benchmarkDimensions.1.items.3", "section": "benchmarks", "kind": "numeric-record", "critical": true, "text": "GPT-5.5 — score: 85.6", "sourceIds": [ "openai-gpt56", "xai-grok45", "moonshot-kimi-k3", "zhipu-glm52", "alibaba-qwen37", "anthropic-claude-3-7-sonnet-system-card", "openai-o3-pro-system-card", "alibaba-qwen-2-5-coder-reasoner-paper", "swe-bench-verified-october-audit" ], "citationOrigin": "explicit-claim-map", "evidenceMode": "registry-review", "evidenceNote": null, "numericValues": [ "85.6" ], "isNumeric": true, "citationCount": 9, "validSourceIds": [ "openai-gpt56", "xai-grok45", "moonshot-kimi-k3", "zhipu-glm52", "alibaba-qwen37", "anthropic-claude-3-7-sonnet-system-card", "openai-o3-pro-system-card", "alibaba-qwen-2-5-coder-reasoner-paper", "swe-bench-verified-october-audit" ], "invalidSourceIds": [], "sourceDetails": [ { "id": "openai-gpt56", "title": "Introducing GPT-5.6", "publisher": "OpenAI", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true }, { "id": "xai-grok45", "title": "Grok 4.5", "publisher": "xAI", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true }, { "id": "moonshot-kimi-k3", "title": "Kimi API platform — Kimi K3 model and pricing", "publisher": "Moonshot AI", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true }, { "id": "zhipu-glm52", "title": "Zhipu BigModel — GLM-5.2 model overview", "publisher": "Zhipu AI", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true }, { "id": "alibaba-qwen37", "title": "Alibaba Cloud Model Studio — Qwen flagship models", "publisher": "Alibaba Cloud", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true }, { "id": "anthropic-claude-3-7-sonnet-system-card", "title": "Anthropic Claude 3.7 Sonnet: Hybrid Reasoning Architecture, Extended Thinking Token Scaling, and Sub-Pixel Computer Use v2", "publisher": "Anthropic", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "openai-o3-pro-system-card", "title": "OpenAI o3-pro System Card: Frontier Reasoning, Test-Time Tree Search with Native Multi-Step Tool Execution, and Autonomous Coding Evaluations", "publisher": "OpenAI", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "alibaba-qwen-2-5-coder-reasoner-paper", "title": "Qwen 2.5-Coder-32B-Reasoner: Open-Weight Test-Time Search Code Agent with Step-Level Process Reward Supervision", "publisher": "Alibaba Cloud & Qwen Team", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "swe-bench-verified-october-audit", "title": "SWE-bench Verified & Terminal-Bench October 2026 Frontier Coding Leaderboard: Test-Time Reasoning vs. Open-Weight Architectures", "publisher": "CodeEval Foundation / SWE-bench Consortium", "grade": "A", "score": 97, "freshnessStatus": "current", "critical": true } ], "bestGrade": "A", "worstGrade": "B", "coverage": { "status": "supported", "reason": "At least one valid source explicitly supports this claim." } }, { "id": "claim:benchmarkDimensions.1.items.4", "path": "benchmarkDimensions.1.items.4", "section": "benchmarks", "kind": "numeric-record", "critical": true, "text": "GPT-5.6 Luna — score: 84.7", "sourceIds": [ "openai-gpt56", "xai-grok45", "moonshot-kimi-k3", "zhipu-glm52", "alibaba-qwen37", "anthropic-claude-3-7-sonnet-system-card", "openai-o3-pro-system-card", "alibaba-qwen-2-5-coder-reasoner-paper", "swe-bench-verified-october-audit" ], "citationOrigin": "explicit-claim-map", "evidenceMode": "registry-review", "evidenceNote": null, "numericValues": [ "84.7" ], "isNumeric": true, "citationCount": 9, "validSourceIds": [ "openai-gpt56", "xai-grok45", "moonshot-kimi-k3", "zhipu-glm52", "alibaba-qwen37", "anthropic-claude-3-7-sonnet-system-card", "openai-o3-pro-system-card", "alibaba-qwen-2-5-coder-reasoner-paper", "swe-bench-verified-october-audit" ], "invalidSourceIds": [], "sourceDetails": [ { "id": "openai-gpt56", "title": "Introducing GPT-5.6", "publisher": "OpenAI", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true }, { "id": "xai-grok45", "title": "Grok 4.5", "publisher": "xAI", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true }, { "id": "moonshot-kimi-k3", "title": "Kimi API platform — Kimi K3 model and pricing", "publisher": "Moonshot AI", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true }, { "id": "zhipu-glm52", "title": "Zhipu BigModel — GLM-5.2 model overview", "publisher": "Zhipu AI", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true }, { "id": "alibaba-qwen37", "title": "Alibaba Cloud Model Studio — Qwen flagship models", "publisher": "Alibaba Cloud", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true }, { "id": "anthropic-claude-3-7-sonnet-system-card", "title": "Anthropic Claude 3.7 Sonnet: Hybrid Reasoning Architecture, Extended Thinking Token Scaling, and Sub-Pixel Computer Use v2", "publisher": "Anthropic", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "openai-o3-pro-system-card", "title": "OpenAI o3-pro System Card: Frontier Reasoning, Test-Time Tree Search with Native Multi-Step Tool Execution, and Autonomous Coding Evaluations", "publisher": "OpenAI", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "alibaba-qwen-2-5-coder-reasoner-paper", "title": "Qwen 2.5-Coder-32B-Reasoner: Open-Weight Test-Time Search Code Agent with Step-Level Process Reward Supervision", "publisher": "Alibaba Cloud & Qwen Team", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "swe-bench-verified-october-audit", "title": "SWE-bench Verified & Terminal-Bench October 2026 Frontier Coding Leaderboard: Test-Time Reasoning vs. Open-Weight Architectures", "publisher": "CodeEval Foundation / SWE-bench Consortium", "grade": "A", "score": 97, "freshnessStatus": "current", "critical": true } ], "bestGrade": "A", "worstGrade": "B", "coverage": { "status": "supported", "reason": "At least one valid source explicitly supports this claim." } }, { "id": "claim:benchmarkDimensions.1.items.5", "path": "benchmarkDimensions.1.items.5", "section": "benchmarks", "kind": "numeric-record", "critical": true, "text": "Claude Fable 5 — score: 83.1", "sourceIds": [ "openai-gpt56", "xai-grok45", "moonshot-kimi-k3", "zhipu-glm52", "alibaba-qwen37", "anthropic-claude-3-7-sonnet-system-card", "openai-o3-pro-system-card", "alibaba-qwen-2-5-coder-reasoner-paper", "swe-bench-verified-october-audit" ], "citationOrigin": "explicit-claim-map", "evidenceMode": "registry-review", "evidenceNote": null, "numericValues": [ "83.1" ], "isNumeric": true, "citationCount": 9, "validSourceIds": [ "openai-gpt56", "xai-grok45", "moonshot-kimi-k3", "zhipu-glm52", "alibaba-qwen37", "anthropic-claude-3-7-sonnet-system-card", "openai-o3-pro-system-card", "alibaba-qwen-2-5-coder-reasoner-paper", "swe-bench-verified-october-audit" ], "invalidSourceIds": [], "sourceDetails": [ { "id": "openai-gpt56", "title": "Introducing GPT-5.6", "publisher": "OpenAI", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true }, { "id": "xai-grok45", "title": "Grok 4.5", "publisher": "xAI", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true }, { "id": "moonshot-kimi-k3", "title": "Kimi API platform — Kimi K3 model and pricing", "publisher": "Moonshot AI", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true }, { "id": "zhipu-glm52", "title": "Zhipu BigModel — GLM-5.2 model overview", "publisher": "Zhipu AI", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true }, { "id": "alibaba-qwen37", "title": "Alibaba Cloud Model Studio — Qwen flagship models", "publisher": "Alibaba Cloud", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true }, { "id": "anthropic-claude-3-7-sonnet-system-card", "title": "Anthropic Claude 3.7 Sonnet: Hybrid Reasoning Architecture, Extended Thinking Token Scaling, and Sub-Pixel Computer Use v2", "publisher": "Anthropic", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "openai-o3-pro-system-card", "title": "OpenAI o3-pro System Card: Frontier Reasoning, Test-Time Tree Search with Native Multi-Step Tool Execution, and Autonomous Coding Evaluations", "publisher": "OpenAI", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "alibaba-qwen-2-5-coder-reasoner-paper", "title": "Qwen 2.5-Coder-32B-Reasoner: Open-Weight Test-Time Search Code Agent with Step-Level Process Reward Supervision", "publisher": "Alibaba Cloud & Qwen Team", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "swe-bench-verified-october-audit", "title": "SWE-bench Verified & Terminal-Bench October 2026 Frontier Coding Leaderboard: Test-Time Reasoning vs. Open-Weight Architectures", "publisher": "CodeEval Foundation / SWE-bench Consortium", "grade": "A", "score": 97, "freshnessStatus": "current", "critical": true } ], "bestGrade": "A", "worstGrade": "B", "coverage": { "status": "supported", "reason": "At least one valid source explicitly supports this claim." } }, { "id": "claim:benchmarkDimensions.1.items.6", "path": "benchmarkDimensions.1.items.6", "section": "benchmarks", "kind": "numeric-record", "critical": true, "text": "Claude Opus 4.8 — score: 78.9", "sourceIds": [ "openai-gpt56", "xai-grok45", "moonshot-kimi-k3", "zhipu-glm52", "alibaba-qwen37", "anthropic-claude-3-7-sonnet-system-card", "openai-o3-pro-system-card", "alibaba-qwen-2-5-coder-reasoner-paper", "swe-bench-verified-october-audit" ], "citationOrigin": "explicit-claim-map", "evidenceMode": "registry-review", "evidenceNote": null, "numericValues": [ "78.9" ], "isNumeric": true, "citationCount": 9, "validSourceIds": [ "openai-gpt56", "xai-grok45", "moonshot-kimi-k3", "zhipu-glm52", "alibaba-qwen37", "anthropic-claude-3-7-sonnet-system-card", "openai-o3-pro-system-card", "alibaba-qwen-2-5-coder-reasoner-paper", "swe-bench-verified-october-audit" ], "invalidSourceIds": [], "sourceDetails": [ { "id": "openai-gpt56", "title": "Introducing GPT-5.6", "publisher": "OpenAI", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true }, { "id": "xai-grok45", "title": "Grok 4.5", "publisher": "xAI", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true }, { "id": "moonshot-kimi-k3", "title": "Kimi API platform — Kimi K3 model and pricing", "publisher": "Moonshot AI", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true }, { "id": "zhipu-glm52", "title": "Zhipu BigModel — GLM-5.2 model overview", "publisher": "Zhipu AI", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true }, { "id": "alibaba-qwen37", "title": "Alibaba Cloud Model Studio — Qwen flagship models", "publisher": "Alibaba Cloud", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true }, { "id": "anthropic-claude-3-7-sonnet-system-card", "title": "Anthropic Claude 3.7 Sonnet: Hybrid Reasoning Architecture, Extended Thinking Token Scaling, and Sub-Pixel Computer Use v2", "publisher": "Anthropic", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "openai-o3-pro-system-card", "title": "OpenAI o3-pro System Card: Frontier Reasoning, Test-Time Tree Search with Native Multi-Step Tool Execution, and Autonomous Coding Evaluations", "publisher": "OpenAI", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "alibaba-qwen-2-5-coder-reasoner-paper", "title": "Qwen 2.5-Coder-32B-Reasoner: Open-Weight Test-Time Search Code Agent with Step-Level Process Reward Supervision", "publisher": "Alibaba Cloud & Qwen Team", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "swe-bench-verified-october-audit", "title": "SWE-bench Verified & Terminal-Bench October 2026 Frontier Coding Leaderboard: Test-Time Reasoning vs. Open-Weight Architectures", "publisher": "CodeEval Foundation / SWE-bench Consortium", "grade": "A", "score": 97, "freshnessStatus": "current", "critical": true } ], "bestGrade": "A", "worstGrade": "B", "coverage": { "status": "supported", "reason": "At least one valid source explicitly supports this claim." } }, { "id": "claim:benchmarkDimensions.1.items.7", "path": "benchmarkDimensions.1.items.7", "section": "benchmarks", "kind": "numeric-record", "critical": true, "text": "Gemini 3.1 Pro Preview — score: 70.7", "sourceIds": [ "openai-gpt56", "xai-grok45", "moonshot-kimi-k3", "zhipu-glm52", "alibaba-qwen37", "anthropic-claude-3-7-sonnet-system-card", "openai-o3-pro-system-card", "alibaba-qwen-2-5-coder-reasoner-paper", "swe-bench-verified-october-audit" ], "citationOrigin": "explicit-claim-map", "evidenceMode": "registry-review", "evidenceNote": null, "numericValues": [ "70.7" ], "isNumeric": true, "citationCount": 9, "validSourceIds": [ "openai-gpt56", "xai-grok45", "moonshot-kimi-k3", "zhipu-glm52", "alibaba-qwen37", "anthropic-claude-3-7-sonnet-system-card", "openai-o3-pro-system-card", "alibaba-qwen-2-5-coder-reasoner-paper", "swe-bench-verified-october-audit" ], "invalidSourceIds": [], "sourceDetails": [ { "id": "openai-gpt56", "title": "Introducing GPT-5.6", "publisher": "OpenAI", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true }, { "id": "xai-grok45", "title": "Grok 4.5", "publisher": "xAI", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true }, { "id": "moonshot-kimi-k3", "title": "Kimi API platform — Kimi K3 model and pricing", "publisher": "Moonshot AI", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true }, { "id": "zhipu-glm52", "title": "Zhipu BigModel — GLM-5.2 model overview", "publisher": "Zhipu AI", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true }, { "id": "alibaba-qwen37", "title": "Alibaba Cloud Model Studio — Qwen flagship models", "publisher": "Alibaba Cloud", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true }, { "id": "anthropic-claude-3-7-sonnet-system-card", "title": "Anthropic Claude 3.7 Sonnet: Hybrid Reasoning Architecture, Extended Thinking Token Scaling, and Sub-Pixel Computer Use v2", "publisher": "Anthropic", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "openai-o3-pro-system-card", "title": "OpenAI o3-pro System Card: Frontier Reasoning, Test-Time Tree Search with Native Multi-Step Tool Execution, and Autonomous Coding Evaluations", "publisher": "OpenAI", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "alibaba-qwen-2-5-coder-reasoner-paper", "title": "Qwen 2.5-Coder-32B-Reasoner: Open-Weight Test-Time Search Code Agent with Step-Level Process Reward Supervision", "publisher": "Alibaba Cloud & Qwen Team", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "swe-bench-verified-october-audit", "title": "SWE-bench Verified & Terminal-Bench October 2026 Frontier Coding Leaderboard: Test-Time Reasoning vs. Open-Weight Architectures", "publisher": "CodeEval Foundation / SWE-bench Consortium", "grade": "A", "score": 97, "freshnessStatus": "current", "critical": true } ], "bestGrade": "A", "worstGrade": "B", "coverage": { "status": "supported", "reason": "At least one valid source explicitly supports this claim." } }, { "id": "claim:benchmarkDimensions.2.status", "path": "benchmarkDimensions.2.status", "section": "benchmarks", "kind": "verification-claim", "critical": true, "text": "Mathematics and science: Framework active; no forced composite", "sourceIds": [ "openai-gpt56", "moonshot-kimi-k3" ], "citationOrigin": "explicit-claim-map", "evidenceMode": "registry-review", "evidenceNote": null, "numericValues": [], "isNumeric": false, "citationCount": 2, "validSourceIds": [ "openai-gpt56", "moonshot-kimi-k3" ], "invalidSourceIds": [], "sourceDetails": [ { "id": "openai-gpt56", "title": "Introducing GPT-5.6", "publisher": "OpenAI", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true }, { "id": "moonshot-kimi-k3", "title": "Kimi API platform — Kimi K3 model and pricing", "publisher": "Moonshot AI", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true } ], "bestGrade": "B", "worstGrade": "B", "coverage": { "status": "supported", "reason": "At least one valid source explicitly supports this claim." } }, { "id": "claim:benchmarkDimensions.2.items.0", "path": "benchmarkDimensions.2.items.0", "section": "benchmarks", "kind": "numeric-record", "critical": true, "text": "GPT-5.6 Sol — score: 94.6", "sourceIds": [ "openai-gpt56", "moonshot-kimi-k3" ], "citationOrigin": "explicit-claim-map", "evidenceMode": "registry-review", "evidenceNote": null, "numericValues": [ "94.6" ], "isNumeric": true, "citationCount": 2, "validSourceIds": [ "openai-gpt56", "moonshot-kimi-k3" ], "invalidSourceIds": [], "sourceDetails": [ { "id": "openai-gpt56", "title": "Introducing GPT-5.6", "publisher": "OpenAI", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true }, { "id": "moonshot-kimi-k3", "title": "Kimi API platform — Kimi K3 model and pricing", "publisher": "Moonshot AI", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true } ], "bestGrade": "B", "worstGrade": "B", "coverage": { "status": "supported", "reason": "At least one valid source explicitly supports this claim." } }, { "id": "claim:benchmarkDimensions.2.items.1", "path": "benchmarkDimensions.2.items.1", "section": "benchmarks", "kind": "numeric-record", "critical": true, "text": "Claude Mythos Preview — score: 94.6", "sourceIds": [ "openai-gpt56", "moonshot-kimi-k3" ], "citationOrigin": "explicit-claim-map", "evidenceMode": "registry-review", "evidenceNote": null, "numericValues": [ "94.6" ], "isNumeric": true, "citationCount": 2, "validSourceIds": [ "openai-gpt56", "moonshot-kimi-k3" ], "invalidSourceIds": [], "sourceDetails": [ { "id": "openai-gpt56", "title": "Introducing GPT-5.6", "publisher": "OpenAI", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true }, { "id": "moonshot-kimi-k3", "title": "Kimi API platform — Kimi K3 model and pricing", "publisher": "Moonshot AI", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true } ], "bestGrade": "B", "worstGrade": "B", "coverage": { "status": "supported", "reason": "At least one valid source explicitly supports this claim." } }, { "id": "claim:benchmarkDimensions.2.items.2", "path": "benchmarkDimensions.2.items.2", "section": "benchmarks", "kind": "numeric-record", "critical": true, "text": "Gemini 3.1 Pro Preview — score: 94.3", "sourceIds": [ "openai-gpt56", "moonshot-kimi-k3" ], "citationOrigin": "explicit-claim-map", "evidenceMode": "registry-review", "evidenceNote": null, "numericValues": [ "94.3" ], "isNumeric": true, "citationCount": 2, "validSourceIds": [ "openai-gpt56", "moonshot-kimi-k3" ], "invalidSourceIds": [], "sourceDetails": [ { "id": "openai-gpt56", "title": "Introducing GPT-5.6", "publisher": "OpenAI", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true }, { "id": "moonshot-kimi-k3", "title": "Kimi API platform — Kimi K3 model and pricing", "publisher": "Moonshot AI", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true } ], "bestGrade": "B", "worstGrade": "B", "coverage": { "status": "supported", "reason": "At least one valid source explicitly supports this claim." } }, { "id": "claim:benchmarkDimensions.2.items.3", "path": "benchmarkDimensions.2.items.3", "section": "benchmarks", "kind": "numeric-record", "critical": true, "text": "Claude Mythos 5 — score: 94.1", "sourceIds": [ "openai-gpt56", "moonshot-kimi-k3" ], "citationOrigin": "explicit-claim-map", "evidenceMode": "registry-review", "evidenceNote": null, "numericValues": [ "94.1" ], "isNumeric": true, "citationCount": 2, "validSourceIds": [ "openai-gpt56", "moonshot-kimi-k3" ], "invalidSourceIds": [], "sourceDetails": [ { "id": "openai-gpt56", "title": "Introducing GPT-5.6", "publisher": "OpenAI", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true }, { "id": "moonshot-kimi-k3", "title": "Kimi API platform — Kimi K3 model and pricing", "publisher": "Moonshot AI", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true } ], "bestGrade": "B", "worstGrade": "B", "coverage": { "status": "supported", "reason": "At least one valid source explicitly supports this claim." } }, { "id": "claim:benchmarkDimensions.2.items.4", "path": "benchmarkDimensions.2.items.4", "section": "benchmarks", "kind": "numeric-record", "critical": true, "text": "GPT-5.5 — score: 93.6", "sourceIds": [ "openai-gpt56", "moonshot-kimi-k3" ], "citationOrigin": "explicit-claim-map", "evidenceMode": "registry-review", "evidenceNote": null, "numericValues": [ "93.6" ], "isNumeric": true, "citationCount": 2, "validSourceIds": [ "openai-gpt56", "moonshot-kimi-k3" ], "invalidSourceIds": [], "sourceDetails": [ { "id": "openai-gpt56", "title": "Introducing GPT-5.6", "publisher": "OpenAI", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true }, { "id": "moonshot-kimi-k3", "title": "Kimi API platform — Kimi K3 model and pricing", "publisher": "Moonshot AI", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true } ], "bestGrade": "B", "worstGrade": "B", "coverage": { "status": "supported", "reason": "At least one valid source explicitly supports this claim." } }, { "id": "claim:benchmarkDimensions.2.items.5", "path": "benchmarkDimensions.2.items.5", "section": "benchmarks", "kind": "numeric-record", "critical": true, "text": "GPT-5.6 Terra — score: 92.9", "sourceIds": [ "openai-gpt56", "moonshot-kimi-k3" ], "citationOrigin": "explicit-claim-map", "evidenceMode": "registry-review", "evidenceNote": null, "numericValues": [ "92.9" ], "isNumeric": true, "citationCount": 2, "validSourceIds": [ "openai-gpt56", "moonshot-kimi-k3" ], "invalidSourceIds": [], "sourceDetails": [ { "id": "openai-gpt56", "title": "Introducing GPT-5.6", "publisher": "OpenAI", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true }, { "id": "moonshot-kimi-k3", "title": "Kimi API platform — Kimi K3 model and pricing", "publisher": "Moonshot AI", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true } ], "bestGrade": "B", "worstGrade": "B", "coverage": { "status": "supported", "reason": "At least one valid source explicitly supports this claim." } }, { "id": "claim:benchmarkDimensions.2.items.6", "path": "benchmarkDimensions.2.items.6", "section": "benchmarks", "kind": "numeric-record", "critical": true, "text": "Claude Fable 5 — score: 92.6", "sourceIds": [ "openai-gpt56", "moonshot-kimi-k3" ], "citationOrigin": "explicit-claim-map", "evidenceMode": "registry-review", "evidenceNote": null, "numericValues": [ "92.6" ], "isNumeric": true, "citationCount": 2, "validSourceIds": [ "openai-gpt56", "moonshot-kimi-k3" ], "invalidSourceIds": [], "sourceDetails": [ { "id": "openai-gpt56", "title": "Introducing GPT-5.6", "publisher": "OpenAI", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true }, { "id": "moonshot-kimi-k3", "title": "Kimi API platform — Kimi K3 model and pricing", "publisher": "Moonshot AI", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true } ], "bestGrade": "B", "worstGrade": "B", "coverage": { "status": "supported", "reason": "At least one valid source explicitly supports this claim." } }, { "id": "claim:benchmarkDimensions.2.items.7", "path": "benchmarkDimensions.2.items.7", "section": "benchmarks", "kind": "numeric-record", "critical": true, "text": "GPT-5.6 Luna — score: 92.3", "sourceIds": [ "openai-gpt56", "moonshot-kimi-k3" ], "citationOrigin": "explicit-claim-map", "evidenceMode": "registry-review", "evidenceNote": null, "numericValues": [ "92.3" ], "isNumeric": true, "citationCount": 2, "validSourceIds": [ "openai-gpt56", "moonshot-kimi-k3" ], "invalidSourceIds": [], "sourceDetails": [ { "id": "openai-gpt56", "title": "Introducing GPT-5.6", "publisher": "OpenAI", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true }, { "id": "moonshot-kimi-k3", "title": "Kimi API platform — Kimi K3 model and pricing", "publisher": "Moonshot AI", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true } ], "bestGrade": "B", "worstGrade": "B", "coverage": { "status": "supported", "reason": "At least one valid source explicitly supports this claim." } }, { "id": "claim:benchmarkDimensions.3.status", "path": "benchmarkDimensions.3.status", "section": "benchmarks", "kind": "verification-claim", "critical": true, "text": "Image, document and visual understanding: Provider signals tracked separately; no new comparable July 22 evaluation verified", "sourceIds": [ "openai-gpt56", "google-gemini35", "zhipu-glm-ocr", "baidu-ernie50", "alibaba-qwen37" ], "citationOrigin": "explicit-claim-map", "evidenceMode": "registry-review", "evidenceNote": null, "numericValues": [], "isNumeric": false, "citationCount": 5, "validSourceIds": [ "openai-gpt56", "google-gemini35", "zhipu-glm-ocr", "baidu-ernie50", "alibaba-qwen37" ], "invalidSourceIds": [], "sourceDetails": [ { "id": "openai-gpt56", "title": "Introducing GPT-5.6", "publisher": "OpenAI", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true }, { "id": "google-gemini35", "title": "Gemini 3.5", "publisher": "Google", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true }, { "id": "zhipu-glm-ocr", "title": "Zhipu BigModel — GLM-OCR", "publisher": "Zhipu AI", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "baidu-ernie50", "title": "Baidu Qianfan — ERNIE 5.0 model list and international price", "publisher": "Baidu AI Cloud", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true }, { "id": "alibaba-qwen37", "title": "Alibaba Cloud Model Studio — Qwen flagship models", "publisher": "Alibaba Cloud", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true } ], "bestGrade": "A", "worstGrade": "B", "coverage": { "status": "supported", "reason": "At least one valid source explicitly supports this claim." } }, { "id": "claim:benchmarkDimensions.3.items.0", "path": "benchmarkDimensions.3.items.0", "section": "benchmarks", "kind": "numeric-record", "critical": true, "text": "GPT-5.6 Sol — score: 83", "sourceIds": [ "openai-gpt56", "google-gemini35", "zhipu-glm-ocr", "baidu-ernie50", "alibaba-qwen37" ], "citationOrigin": "explicit-claim-map", "evidenceMode": "registry-review", "evidenceNote": null, "numericValues": [ "83" ], "isNumeric": true, "citationCount": 5, "validSourceIds": [ "openai-gpt56", "google-gemini35", "zhipu-glm-ocr", "baidu-ernie50", "alibaba-qwen37" ], "invalidSourceIds": [], "sourceDetails": [ { "id": "openai-gpt56", "title": "Introducing GPT-5.6", "publisher": "OpenAI", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true }, { "id": "google-gemini35", "title": "Gemini 3.5", "publisher": "Google", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true }, { "id": "zhipu-glm-ocr", "title": "Zhipu BigModel — GLM-OCR", "publisher": "Zhipu AI", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "baidu-ernie50", "title": "Baidu Qianfan — ERNIE 5.0 model list and international price", "publisher": "Baidu AI Cloud", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true }, { "id": "alibaba-qwen37", "title": "Alibaba Cloud Model Studio — Qwen flagship models", "publisher": "Alibaba Cloud", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true } ], "bestGrade": "A", "worstGrade": "B", "coverage": { "status": "supported", "reason": "At least one valid source explicitly supports this claim." } }, { "id": "claim:benchmarkDimensions.3.items.1", "path": "benchmarkDimensions.3.items.1", "section": "benchmarks", "kind": "numeric-record", "critical": true, "text": "GPT-5.5 — score: 81.2", "sourceIds": [ "openai-gpt56", "google-gemini35", "zhipu-glm-ocr", "baidu-ernie50", "alibaba-qwen37" ], "citationOrigin": "explicit-claim-map", "evidenceMode": "registry-review", "evidenceNote": null, "numericValues": [ "81.2" ], "isNumeric": true, "citationCount": 5, "validSourceIds": [ "openai-gpt56", "google-gemini35", "zhipu-glm-ocr", "baidu-ernie50", "alibaba-qwen37" ], "invalidSourceIds": [], "sourceDetails": [ { "id": "openai-gpt56", "title": "Introducing GPT-5.6", "publisher": "OpenAI", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true }, { "id": "google-gemini35", "title": "Gemini 3.5", "publisher": "Google", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true }, { "id": "zhipu-glm-ocr", "title": "Zhipu BigModel — GLM-OCR", "publisher": "Zhipu AI", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "baidu-ernie50", "title": "Baidu Qianfan — ERNIE 5.0 model list and international price", "publisher": "Baidu AI Cloud", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true }, { "id": "alibaba-qwen37", "title": "Alibaba Cloud Model Studio — Qwen flagship models", "publisher": "Alibaba Cloud", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true } ], "bestGrade": "A", "worstGrade": "B", "coverage": { "status": "supported", "reason": "At least one valid source explicitly supports this claim." } }, { "id": "claim:benchmarkDimensions.3.items.2", "path": "benchmarkDimensions.3.items.2", "section": "benchmarks", "kind": "numeric-record", "critical": true, "text": "GPT-5.6 Terra — score: 80.7", "sourceIds": [ "openai-gpt56", "google-gemini35", "zhipu-glm-ocr", "baidu-ernie50", "alibaba-qwen37" ], "citationOrigin": "explicit-claim-map", "evidenceMode": "registry-review", "evidenceNote": null, "numericValues": [ "80.7" ], "isNumeric": true, "citationCount": 5, "validSourceIds": [ "openai-gpt56", "google-gemini35", "zhipu-glm-ocr", "baidu-ernie50", "alibaba-qwen37" ], "invalidSourceIds": [], "sourceDetails": [ { "id": "openai-gpt56", "title": "Introducing GPT-5.6", "publisher": "OpenAI", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true }, { "id": "google-gemini35", "title": "Gemini 3.5", "publisher": "Google", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true }, { "id": "zhipu-glm-ocr", "title": "Zhipu BigModel — GLM-OCR", "publisher": "Zhipu AI", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "baidu-ernie50", "title": "Baidu Qianfan — ERNIE 5.0 model list and international price", "publisher": "Baidu AI Cloud", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true }, { "id": "alibaba-qwen37", "title": "Alibaba Cloud Model Studio — Qwen flagship models", "publisher": "Alibaba Cloud", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true } ], "bestGrade": "A", "worstGrade": "B", "coverage": { "status": "supported", "reason": "At least one valid source explicitly supports this claim." } }, { "id": "claim:benchmarkDimensions.3.items.3", "path": "benchmarkDimensions.3.items.3", "section": "benchmarks", "kind": "numeric-record", "critical": true, "text": "Gemini 3.1 Pro Preview — score: 80.5", "sourceIds": [ "openai-gpt56", "google-gemini35", "zhipu-glm-ocr", "baidu-ernie50", "alibaba-qwen37" ], "citationOrigin": "explicit-claim-map", "evidenceMode": "registry-review", "evidenceNote": null, "numericValues": [ "80.5" ], "isNumeric": true, "citationCount": 5, "validSourceIds": [ "openai-gpt56", "google-gemini35", "zhipu-glm-ocr", "baidu-ernie50", "alibaba-qwen37" ], "invalidSourceIds": [], "sourceDetails": [ { "id": "openai-gpt56", "title": "Introducing GPT-5.6", "publisher": "OpenAI", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true }, { "id": "google-gemini35", "title": "Gemini 3.5", "publisher": "Google", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true }, { "id": "zhipu-glm-ocr", "title": "Zhipu BigModel — GLM-OCR", "publisher": "Zhipu AI", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "baidu-ernie50", "title": "Baidu Qianfan — ERNIE 5.0 model list and international price", "publisher": "Baidu AI Cloud", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true }, { "id": "alibaba-qwen37", "title": "Alibaba Cloud Model Studio — Qwen flagship models", "publisher": "Alibaba Cloud", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true } ], "bestGrade": "A", "worstGrade": "B", "coverage": { "status": "supported", "reason": "At least one valid source explicitly supports this claim." } }, { "id": "claim:benchmarkDimensions.3.items.4", "path": "benchmarkDimensions.3.items.4", "section": "benchmarks", "kind": "numeric-record", "critical": true, "text": "GPT-5.6 Luna — score: 78.4", "sourceIds": [ "openai-gpt56", "google-gemini35", "zhipu-glm-ocr", "baidu-ernie50", "alibaba-qwen37" ], "citationOrigin": "explicit-claim-map", "evidenceMode": "registry-review", "evidenceNote": null, "numericValues": [ "78.4" ], "isNumeric": true, "citationCount": 5, "validSourceIds": [ "openai-gpt56", "google-gemini35", "zhipu-glm-ocr", "baidu-ernie50", "alibaba-qwen37" ], "invalidSourceIds": [], "sourceDetails": [ { "id": "openai-gpt56", "title": "Introducing GPT-5.6", "publisher": "OpenAI", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true }, { "id": "google-gemini35", "title": "Gemini 3.5", "publisher": "Google", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true }, { "id": "zhipu-glm-ocr", "title": "Zhipu BigModel — GLM-OCR", "publisher": "Zhipu AI", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "baidu-ernie50", "title": "Baidu Qianfan — ERNIE 5.0 model list and international price", "publisher": "Baidu AI Cloud", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true }, { "id": "alibaba-qwen37", "title": "Alibaba Cloud Model Studio — Qwen flagship models", "publisher": "Alibaba Cloud", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true } ], "bestGrade": "A", "worstGrade": "B", "coverage": { "status": "supported", "reason": "At least one valid source explicitly supports this claim." } }, { "id": "claim:benchmarkDimensions.4.status", "path": "benchmarkDimensions.4.status", "section": "benchmarks", "kind": "verification-claim", "critical": true, "text": "Image generation and editing: Human preference and prompt fidelity tracked; no new comparable July 22 evaluation verified", "sourceIds": [ "zhipu-glm52", "alibaba-qwen37", "baidu-ernie50" ], "citationOrigin": "explicit-claim-map", "evidenceMode": "registry-review", "evidenceNote": null, "numericValues": [], "isNumeric": false, "citationCount": 3, "validSourceIds": [ "zhipu-glm52", "alibaba-qwen37", "baidu-ernie50" ], "invalidSourceIds": [], "sourceDetails": [ { "id": "zhipu-glm52", "title": "Zhipu BigModel — GLM-5.2 model overview", "publisher": "Zhipu AI", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true }, { "id": "alibaba-qwen37", "title": "Alibaba Cloud Model Studio — Qwen flagship models", "publisher": "Alibaba Cloud", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true }, { "id": "baidu-ernie50", "title": "Baidu Qianfan — ERNIE 5.0 model list and international price", "publisher": "Baidu AI Cloud", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true } ], "bestGrade": "B", "worstGrade": "B", "coverage": { "status": "supported", "reason": "At least one valid source explicitly supports this claim." } }, { "id": "claim:benchmarkDimensions.5.status", "path": "benchmarkDimensions.5.status", "section": "benchmarks", "kind": "verification-claim", "critical": true, "text": "Video, speech and audio: Separate modality scorecards; no new comparable July 22 evaluation verified", "sourceIds": [ "google-gemini35", "baidu-ernie50", "alibaba-qwen37" ], "citationOrigin": "explicit-claim-map", "evidenceMode": "registry-review", "evidenceNote": null, "numericValues": [], "isNumeric": false, "citationCount": 3, "validSourceIds": [ "google-gemini35", "baidu-ernie50", "alibaba-qwen37" ], "invalidSourceIds": [], "sourceDetails": [ { "id": "google-gemini35", "title": "Gemini 3.5", "publisher": "Google", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true }, { "id": "baidu-ernie50", "title": "Baidu Qianfan — ERNIE 5.0 model list and international price", "publisher": "Baidu AI Cloud", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true }, { "id": "alibaba-qwen37", "title": "Alibaba Cloud Model Studio — Qwen flagship models", "publisher": "Alibaba Cloud", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true } ], "bestGrade": "B", "worstGrade": "B", "coverage": { "status": "supported", "reason": "At least one valid source explicitly supports this claim." } }, { "id": "claim:benchmarkDimensions.6.status", "path": "benchmarkDimensions.6.status", "section": "benchmarks", "kind": "verification-claim", "critical": true, "text": "Agents, tools and web research: Operational success metrics prioritized; no new comparable July 22 evaluation verified", "sourceIds": [ "openai-gpt56", "google-gemini35", "moonshot-kimi-k3", "zhipu-glm52", "anthropic-claude-3-7-sonnet-system-card", "openai-o3-pro-system-card", "openai-operator-sandbox-architecture", "nist-aisi-agentic-privilege-escalation-bench", "anthropic-claude-3-8-system-card", "deepseek-v3-5-architecture-report", "anthropic-claude-motion-usage-policy", "meta-muse-windows-native" ], "citationOrigin": "explicit-claim-map", "evidenceMode": "registry-review", "evidenceNote": null, "numericValues": [], "isNumeric": false, "citationCount": 12, "validSourceIds": [ "openai-gpt56", "google-gemini35", "moonshot-kimi-k3", "zhipu-glm52", "anthropic-claude-3-7-sonnet-system-card", "openai-o3-pro-system-card", "openai-operator-sandbox-architecture", "nist-aisi-agentic-privilege-escalation-bench", "anthropic-claude-3-8-system-card", "deepseek-v3-5-architecture-report", "anthropic-claude-motion-usage-policy", "meta-muse-windows-native" ], "invalidSourceIds": [], "sourceDetails": [ { "id": "openai-gpt56", "title": "Introducing GPT-5.6", "publisher": "OpenAI", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true }, { "id": "google-gemini35", "title": "Gemini 3.5", "publisher": "Google", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true }, { "id": "moonshot-kimi-k3", "title": "Kimi API platform — Kimi K3 model and pricing", "publisher": "Moonshot AI", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true }, { "id": "zhipu-glm52", "title": "Zhipu BigModel — GLM-5.2 model overview", "publisher": "Zhipu AI", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true }, { "id": "anthropic-claude-3-7-sonnet-system-card", "title": "Anthropic Claude 3.7 Sonnet: Hybrid Reasoning Architecture, Extended Thinking Token Scaling, and Sub-Pixel Computer Use v2", "publisher": "Anthropic", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "openai-o3-pro-system-card", "title": "OpenAI o3-pro System Card: Frontier Reasoning, Test-Time Tree Search with Native Multi-Step Tool Execution, and Autonomous Coding Evaluations", "publisher": "OpenAI", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "openai-operator-sandbox-architecture", "title": "OpenAI Operator Architecture: MicroVM Hardware Isolation, Permission Gating, and Dual-Loop Verification for Web Agents", "publisher": "OpenAI", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "nist-aisi-agentic-privilege-escalation-bench", "title": "US Artificial Intelligence Safety Institute (NIST AISI): Standardized Test Suite for Agentic Privilege Escalation and MicroVM Containment", "publisher": "National Institute of Standards and Technology (NIST)", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "anthropic-claude-3-8-system-card", "title": "Anthropic Claude 3.8 Architecture & System Card: Computer Use v2.5, POSIX Bash Sandboxing, and SWE-bench Verified Benchmark", "publisher": "Anthropic PBC", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "deepseek-v3-5-architecture-report", "title": "DeepSeek-V3.5 Technical Report: Multi-Head Latent Attention (MLA) Scaling and DualPipe Parallelism across 2,048 Compute Nodes", "publisher": "DeepSeek AI", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "anthropic-claude-motion-usage-policy", "title": "Anthropic Claude Motion Engine, Dashboards Beta, and System Call Containment Policy", "publisher": "Anthropic PBC", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "meta-muse-windows-native", "title": "Meta Muse Desktop Architecture: Native Windows Runtime, Local Memory Buffers, and Cross-Platform Integration", "publisher": "Meta Platforms & Microsoft", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true } ], "bestGrade": "A", "worstGrade": "B", "coverage": { "status": "supported", "reason": "At least one valid source explicitly supports this claim." } }, { "id": "claim:benchmarkDimensions.6.items.0", "path": "benchmarkDimensions.6.items.0", "section": "benchmarks", "kind": "numeric-record", "critical": true, "text": "GPT-5.6 Sol — score: 90.4", "sourceIds": [ "openai-gpt56", "google-gemini35", "moonshot-kimi-k3", "zhipu-glm52", "anthropic-claude-3-7-sonnet-system-card", "openai-o3-pro-system-card", "openai-operator-sandbox-architecture", "nist-aisi-agentic-privilege-escalation-bench", "anthropic-claude-3-8-system-card", "deepseek-v3-5-architecture-report", "anthropic-claude-motion-usage-policy", "meta-muse-windows-native" ], "citationOrigin": "explicit-claim-map", "evidenceMode": "registry-review", "evidenceNote": null, "numericValues": [ "90.4" ], "isNumeric": true, "citationCount": 12, "validSourceIds": [ "openai-gpt56", "google-gemini35", "moonshot-kimi-k3", "zhipu-glm52", "anthropic-claude-3-7-sonnet-system-card", "openai-o3-pro-system-card", "openai-operator-sandbox-architecture", "nist-aisi-agentic-privilege-escalation-bench", "anthropic-claude-3-8-system-card", "deepseek-v3-5-architecture-report", "anthropic-claude-motion-usage-policy", "meta-muse-windows-native" ], "invalidSourceIds": [], "sourceDetails": [ { "id": "openai-gpt56", "title": "Introducing GPT-5.6", "publisher": "OpenAI", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true }, { "id": "google-gemini35", "title": "Gemini 3.5", "publisher": "Google", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true }, { "id": "moonshot-kimi-k3", "title": "Kimi API platform — Kimi K3 model and pricing", "publisher": "Moonshot AI", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true }, { "id": "zhipu-glm52", "title": "Zhipu BigModel — GLM-5.2 model overview", "publisher": "Zhipu AI", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true }, { "id": "anthropic-claude-3-7-sonnet-system-card", "title": "Anthropic Claude 3.7 Sonnet: Hybrid Reasoning Architecture, Extended Thinking Token Scaling, and Sub-Pixel Computer Use v2", "publisher": "Anthropic", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "openai-o3-pro-system-card", "title": "OpenAI o3-pro System Card: Frontier Reasoning, Test-Time Tree Search with Native Multi-Step Tool Execution, and Autonomous Coding Evaluations", "publisher": "OpenAI", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "openai-operator-sandbox-architecture", "title": "OpenAI Operator Architecture: MicroVM Hardware Isolation, Permission Gating, and Dual-Loop Verification for Web Agents", "publisher": "OpenAI", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "nist-aisi-agentic-privilege-escalation-bench", "title": "US Artificial Intelligence Safety Institute (NIST AISI): Standardized Test Suite for Agentic Privilege Escalation and MicroVM Containment", "publisher": "National Institute of Standards and Technology (NIST)", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "anthropic-claude-3-8-system-card", "title": "Anthropic Claude 3.8 Architecture & System Card: Computer Use v2.5, POSIX Bash Sandboxing, and SWE-bench Verified Benchmark", "publisher": "Anthropic PBC", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "deepseek-v3-5-architecture-report", "title": "DeepSeek-V3.5 Technical Report: Multi-Head Latent Attention (MLA) Scaling and DualPipe Parallelism across 2,048 Compute Nodes", "publisher": "DeepSeek AI", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "anthropic-claude-motion-usage-policy", "title": "Anthropic Claude Motion Engine, Dashboards Beta, and System Call Containment Policy", "publisher": "Anthropic PBC", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "meta-muse-windows-native", "title": "Meta Muse Desktop Architecture: Native Windows Runtime, Local Memory Buffers, and Cross-Platform Integration", "publisher": "Meta Platforms & Microsoft", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true } ], "bestGrade": "A", "worstGrade": "B", "coverage": { "status": "supported", "reason": "At least one valid source explicitly supports this claim." } }, { "id": "claim:benchmarkDimensions.6.items.1", "path": "benchmarkDimensions.6.items.1", "section": "benchmarks", "kind": "numeric-record", "critical": true, "text": "Claude Mythos 5 — score: 88", "sourceIds": [ "openai-gpt56", "google-gemini35", "moonshot-kimi-k3", "zhipu-glm52", "anthropic-claude-3-7-sonnet-system-card", "openai-o3-pro-system-card", "openai-operator-sandbox-architecture", "nist-aisi-agentic-privilege-escalation-bench", "anthropic-claude-3-8-system-card", "deepseek-v3-5-architecture-report", "anthropic-claude-motion-usage-policy", "meta-muse-windows-native" ], "citationOrigin": "explicit-claim-map", "evidenceMode": "registry-review", "evidenceNote": null, "numericValues": [ "88" ], "isNumeric": true, "citationCount": 12, "validSourceIds": [ "openai-gpt56", "google-gemini35", "moonshot-kimi-k3", "zhipu-glm52", "anthropic-claude-3-7-sonnet-system-card", "openai-o3-pro-system-card", "openai-operator-sandbox-architecture", "nist-aisi-agentic-privilege-escalation-bench", "anthropic-claude-3-8-system-card", "deepseek-v3-5-architecture-report", "anthropic-claude-motion-usage-policy", "meta-muse-windows-native" ], "invalidSourceIds": [], "sourceDetails": [ { "id": "openai-gpt56", "title": "Introducing GPT-5.6", "publisher": "OpenAI", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true }, { "id": "google-gemini35", "title": "Gemini 3.5", "publisher": "Google", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true }, { "id": "moonshot-kimi-k3", "title": "Kimi API platform — Kimi K3 model and pricing", "publisher": "Moonshot AI", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true }, { "id": "zhipu-glm52", "title": "Zhipu BigModel — GLM-5.2 model overview", "publisher": "Zhipu AI", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true }, { "id": "anthropic-claude-3-7-sonnet-system-card", "title": "Anthropic Claude 3.7 Sonnet: Hybrid Reasoning Architecture, Extended Thinking Token Scaling, and Sub-Pixel Computer Use v2", "publisher": "Anthropic", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "openai-o3-pro-system-card", "title": "OpenAI o3-pro System Card: Frontier Reasoning, Test-Time Tree Search with Native Multi-Step Tool Execution, and Autonomous Coding Evaluations", "publisher": "OpenAI", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "openai-operator-sandbox-architecture", "title": "OpenAI Operator Architecture: MicroVM Hardware Isolation, Permission Gating, and Dual-Loop Verification for Web Agents", "publisher": "OpenAI", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "nist-aisi-agentic-privilege-escalation-bench", "title": "US Artificial Intelligence Safety Institute (NIST AISI): Standardized Test Suite for Agentic Privilege Escalation and MicroVM Containment", "publisher": "National Institute of Standards and Technology (NIST)", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "anthropic-claude-3-8-system-card", "title": "Anthropic Claude 3.8 Architecture & System Card: Computer Use v2.5, POSIX Bash Sandboxing, and SWE-bench Verified Benchmark", "publisher": "Anthropic PBC", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "deepseek-v3-5-architecture-report", "title": "DeepSeek-V3.5 Technical Report: Multi-Head Latent Attention (MLA) Scaling and DualPipe Parallelism across 2,048 Compute Nodes", "publisher": "DeepSeek AI", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "anthropic-claude-motion-usage-policy", "title": "Anthropic Claude Motion Engine, Dashboards Beta, and System Call Containment Policy", "publisher": "Anthropic PBC", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "meta-muse-windows-native", "title": "Meta Muse Desktop Architecture: Native Windows Runtime, Local Memory Buffers, and Cross-Platform Integration", "publisher": "Meta Platforms & Microsoft", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true } ], "bestGrade": "A", "worstGrade": "B", "coverage": { "status": "supported", "reason": "At least one valid source explicitly supports this claim." } }, { "id": "claim:benchmarkDimensions.6.items.2", "path": "benchmarkDimensions.6.items.2", "section": "benchmarks", "kind": "numeric-record", "critical": true, "text": "Claude Mythos Preview — score: 87.9", "sourceIds": [ "openai-gpt56", "google-gemini35", "moonshot-kimi-k3", "zhipu-glm52", "anthropic-claude-3-7-sonnet-system-card", "openai-o3-pro-system-card", "openai-operator-sandbox-architecture", "nist-aisi-agentic-privilege-escalation-bench", "anthropic-claude-3-8-system-card", "deepseek-v3-5-architecture-report", "anthropic-claude-motion-usage-policy", "meta-muse-windows-native" ], "citationOrigin": "explicit-claim-map", "evidenceMode": "registry-review", "evidenceNote": null, "numericValues": [ "87.9" ], "isNumeric": true, "citationCount": 12, "validSourceIds": [ "openai-gpt56", "google-gemini35", "moonshot-kimi-k3", "zhipu-glm52", "anthropic-claude-3-7-sonnet-system-card", "openai-o3-pro-system-card", "openai-operator-sandbox-architecture", "nist-aisi-agentic-privilege-escalation-bench", "anthropic-claude-3-8-system-card", "deepseek-v3-5-architecture-report", "anthropic-claude-motion-usage-policy", "meta-muse-windows-native" ], "invalidSourceIds": [], "sourceDetails": [ { "id": "openai-gpt56", "title": "Introducing GPT-5.6", "publisher": "OpenAI", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true }, { "id": "google-gemini35", "title": "Gemini 3.5", "publisher": "Google", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true }, { "id": "moonshot-kimi-k3", "title": "Kimi API platform — Kimi K3 model and pricing", "publisher": "Moonshot AI", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true }, { "id": "zhipu-glm52", "title": "Zhipu BigModel — GLM-5.2 model overview", "publisher": "Zhipu AI", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true }, { "id": "anthropic-claude-3-7-sonnet-system-card", "title": "Anthropic Claude 3.7 Sonnet: Hybrid Reasoning Architecture, Extended Thinking Token Scaling, and Sub-Pixel Computer Use v2", "publisher": "Anthropic", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "openai-o3-pro-system-card", "title": "OpenAI o3-pro System Card: Frontier Reasoning, Test-Time Tree Search with Native Multi-Step Tool Execution, and Autonomous Coding Evaluations", "publisher": "OpenAI", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "openai-operator-sandbox-architecture", "title": "OpenAI Operator Architecture: MicroVM Hardware Isolation, Permission Gating, and Dual-Loop Verification for Web Agents", "publisher": "OpenAI", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "nist-aisi-agentic-privilege-escalation-bench", "title": "US Artificial Intelligence Safety Institute (NIST AISI): Standardized Test Suite for Agentic Privilege Escalation and MicroVM Containment", "publisher": "National Institute of Standards and Technology (NIST)", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "anthropic-claude-3-8-system-card", "title": "Anthropic Claude 3.8 Architecture & System Card: Computer Use v2.5, POSIX Bash Sandboxing, and SWE-bench Verified Benchmark", "publisher": "Anthropic PBC", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "deepseek-v3-5-architecture-report", "title": "DeepSeek-V3.5 Technical Report: Multi-Head Latent Attention (MLA) Scaling and DualPipe Parallelism across 2,048 Compute Nodes", "publisher": "DeepSeek AI", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "anthropic-claude-motion-usage-policy", "title": "Anthropic Claude Motion Engine, Dashboards Beta, and System Call Containment Policy", "publisher": "Anthropic PBC", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "meta-muse-windows-native", "title": "Meta Muse Desktop Architecture: Native Windows Runtime, Local Memory Buffers, and Cross-Platform Integration", "publisher": "Meta Platforms & Microsoft", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true } ], "bestGrade": "A", "worstGrade": "B", "coverage": { "status": "supported", "reason": "At least one valid source explicitly supports this claim." } }, { "id": "claim:benchmarkDimensions.6.items.3", "path": "benchmarkDimensions.6.items.3", "section": "benchmarks", "kind": "numeric-record", "critical": true, "text": "Claude Opus 4.8 — score: 84.3", "sourceIds": [ "openai-gpt56", "google-gemini35", "moonshot-kimi-k3", "zhipu-glm52", "anthropic-claude-3-7-sonnet-system-card", "openai-o3-pro-system-card", "openai-operator-sandbox-architecture", "nist-aisi-agentic-privilege-escalation-bench", "anthropic-claude-3-8-system-card", "deepseek-v3-5-architecture-report", "anthropic-claude-motion-usage-policy", "meta-muse-windows-native" ], "citationOrigin": "explicit-claim-map", "evidenceMode": "registry-review", "evidenceNote": null, "numericValues": [ "84.3" ], "isNumeric": true, "citationCount": 12, "validSourceIds": [ "openai-gpt56", "google-gemini35", "moonshot-kimi-k3", "zhipu-glm52", "anthropic-claude-3-7-sonnet-system-card", "openai-o3-pro-system-card", "openai-operator-sandbox-architecture", "nist-aisi-agentic-privilege-escalation-bench", "anthropic-claude-3-8-system-card", "deepseek-v3-5-architecture-report", "anthropic-claude-motion-usage-policy", "meta-muse-windows-native" ], "invalidSourceIds": [], "sourceDetails": [ { "id": "openai-gpt56", "title": "Introducing GPT-5.6", "publisher": "OpenAI", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true }, { "id": "google-gemini35", "title": "Gemini 3.5", "publisher": "Google", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true }, { "id": "moonshot-kimi-k3", "title": "Kimi API platform — Kimi K3 model and pricing", "publisher": "Moonshot AI", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true }, { "id": "zhipu-glm52", "title": "Zhipu BigModel — GLM-5.2 model overview", "publisher": "Zhipu AI", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true }, { "id": "anthropic-claude-3-7-sonnet-system-card", "title": "Anthropic Claude 3.7 Sonnet: Hybrid Reasoning Architecture, Extended Thinking Token Scaling, and Sub-Pixel Computer Use v2", "publisher": "Anthropic", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "openai-o3-pro-system-card", "title": "OpenAI o3-pro System Card: Frontier Reasoning, Test-Time Tree Search with Native Multi-Step Tool Execution, and Autonomous Coding Evaluations", "publisher": "OpenAI", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "openai-operator-sandbox-architecture", "title": "OpenAI Operator Architecture: MicroVM Hardware Isolation, Permission Gating, and Dual-Loop Verification for Web Agents", "publisher": "OpenAI", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "nist-aisi-agentic-privilege-escalation-bench", "title": "US Artificial Intelligence Safety Institute (NIST AISI): Standardized Test Suite for Agentic Privilege Escalation and MicroVM Containment", "publisher": "National Institute of Standards and Technology (NIST)", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "anthropic-claude-3-8-system-card", "title": "Anthropic Claude 3.8 Architecture & System Card: Computer Use v2.5, POSIX Bash Sandboxing, and SWE-bench Verified Benchmark", "publisher": "Anthropic PBC", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "deepseek-v3-5-architecture-report", "title": "DeepSeek-V3.5 Technical Report: Multi-Head Latent Attention (MLA) Scaling and DualPipe Parallelism across 2,048 Compute Nodes", "publisher": "DeepSeek AI", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "anthropic-claude-motion-usage-policy", "title": "Anthropic Claude Motion Engine, Dashboards Beta, and System Call Containment Policy", "publisher": "Anthropic PBC", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "meta-muse-windows-native", "title": "Meta Muse Desktop Architecture: Native Windows Runtime, Local Memory Buffers, and Cross-Platform Integration", "publisher": "Meta Platforms & Microsoft", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true } ], "bestGrade": "A", "worstGrade": "B", "coverage": { "status": "supported", "reason": "At least one valid source explicitly supports this claim." } }, { "id": "claim:benchmarkDimensions.6.items.4", "path": "benchmarkDimensions.6.items.4", "section": "benchmarks", "kind": "numeric-record", "critical": true, "text": "GPT-5.5 — score: 84.4", "sourceIds": [ "openai-gpt56", "google-gemini35", "moonshot-kimi-k3", "zhipu-glm52", "anthropic-claude-3-7-sonnet-system-card", "openai-o3-pro-system-card", "openai-operator-sandbox-architecture", "nist-aisi-agentic-privilege-escalation-bench", "anthropic-claude-3-8-system-card", "deepseek-v3-5-architecture-report", "anthropic-claude-motion-usage-policy", "meta-muse-windows-native" ], "citationOrigin": "explicit-claim-map", "evidenceMode": "registry-review", "evidenceNote": null, "numericValues": [ "84.4" ], "isNumeric": true, "citationCount": 12, "validSourceIds": [ "openai-gpt56", "google-gemini35", "moonshot-kimi-k3", "zhipu-glm52", "anthropic-claude-3-7-sonnet-system-card", "openai-o3-pro-system-card", "openai-operator-sandbox-architecture", "nist-aisi-agentic-privilege-escalation-bench", "anthropic-claude-3-8-system-card", "deepseek-v3-5-architecture-report", "anthropic-claude-motion-usage-policy", "meta-muse-windows-native" ], "invalidSourceIds": [], "sourceDetails": [ { "id": "openai-gpt56", "title": "Introducing GPT-5.6", "publisher": "OpenAI", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true }, { "id": "google-gemini35", "title": "Gemini 3.5", "publisher": "Google", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true }, { "id": "moonshot-kimi-k3", "title": "Kimi API platform — Kimi K3 model and pricing", "publisher": "Moonshot AI", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true }, { "id": "zhipu-glm52", "title": "Zhipu BigModel — GLM-5.2 model overview", "publisher": "Zhipu AI", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true }, { "id": "anthropic-claude-3-7-sonnet-system-card", "title": "Anthropic Claude 3.7 Sonnet: Hybrid Reasoning Architecture, Extended Thinking Token Scaling, and Sub-Pixel Computer Use v2", "publisher": "Anthropic", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "openai-o3-pro-system-card", "title": "OpenAI o3-pro System Card: Frontier Reasoning, Test-Time Tree Search with Native Multi-Step Tool Execution, and Autonomous Coding Evaluations", "publisher": "OpenAI", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "openai-operator-sandbox-architecture", "title": "OpenAI Operator Architecture: MicroVM Hardware Isolation, Permission Gating, and Dual-Loop Verification for Web Agents", "publisher": "OpenAI", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "nist-aisi-agentic-privilege-escalation-bench", "title": "US Artificial Intelligence Safety Institute (NIST AISI): Standardized Test Suite for Agentic Privilege Escalation and MicroVM Containment", "publisher": "National Institute of Standards and Technology (NIST)", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "anthropic-claude-3-8-system-card", "title": "Anthropic Claude 3.8 Architecture & System Card: Computer Use v2.5, POSIX Bash Sandboxing, and SWE-bench Verified Benchmark", "publisher": "Anthropic PBC", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "deepseek-v3-5-architecture-report", "title": "DeepSeek-V3.5 Technical Report: Multi-Head Latent Attention (MLA) Scaling and DualPipe Parallelism across 2,048 Compute Nodes", "publisher": "DeepSeek AI", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "anthropic-claude-motion-usage-policy", "title": "Anthropic Claude Motion Engine, Dashboards Beta, and System Call Containment Policy", "publisher": "Anthropic PBC", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "meta-muse-windows-native", "title": "Meta Muse Desktop Architecture: Native Windows Runtime, Local Memory Buffers, and Cross-Platform Integration", "publisher": "Meta Platforms & Microsoft", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true } ], "bestGrade": "A", "worstGrade": "B", "coverage": { "status": "supported", "reason": "At least one valid source explicitly supports this claim." } }, { "id": "claim:benchmarkDimensions.6.items.5", "path": "benchmarkDimensions.6.items.5", "section": "benchmarks", "kind": "numeric-record", "critical": true, "text": "GPT-5.6 Terra — score: 87.5", "sourceIds": [ "openai-gpt56", "google-gemini35", "moonshot-kimi-k3", "zhipu-glm52", "anthropic-claude-3-7-sonnet-system-card", "openai-o3-pro-system-card", "openai-operator-sandbox-architecture", "nist-aisi-agentic-privilege-escalation-bench", "anthropic-claude-3-8-system-card", "deepseek-v3-5-architecture-report", "anthropic-claude-motion-usage-policy", "meta-muse-windows-native" ], "citationOrigin": "explicit-claim-map", "evidenceMode": "registry-review", "evidenceNote": null, "numericValues": [ "87.5" ], "isNumeric": true, "citationCount": 12, "validSourceIds": [ "openai-gpt56", "google-gemini35", "moonshot-kimi-k3", "zhipu-glm52", "anthropic-claude-3-7-sonnet-system-card", "openai-o3-pro-system-card", "openai-operator-sandbox-architecture", "nist-aisi-agentic-privilege-escalation-bench", "anthropic-claude-3-8-system-card", "deepseek-v3-5-architecture-report", "anthropic-claude-motion-usage-policy", "meta-muse-windows-native" ], "invalidSourceIds": [], "sourceDetails": [ { "id": "openai-gpt56", "title": "Introducing GPT-5.6", "publisher": "OpenAI", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true }, { "id": "google-gemini35", "title": "Gemini 3.5", "publisher": "Google", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true }, { "id": "moonshot-kimi-k3", "title": "Kimi API platform — Kimi K3 model and pricing", "publisher": "Moonshot AI", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true }, { "id": "zhipu-glm52", "title": "Zhipu BigModel — GLM-5.2 model overview", "publisher": "Zhipu AI", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true }, { "id": "anthropic-claude-3-7-sonnet-system-card", "title": "Anthropic Claude 3.7 Sonnet: Hybrid Reasoning Architecture, Extended Thinking Token Scaling, and Sub-Pixel Computer Use v2", "publisher": "Anthropic", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "openai-o3-pro-system-card", "title": "OpenAI o3-pro System Card: Frontier Reasoning, Test-Time Tree Search with Native Multi-Step Tool Execution, and Autonomous Coding Evaluations", "publisher": "OpenAI", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "openai-operator-sandbox-architecture", "title": "OpenAI Operator Architecture: MicroVM Hardware Isolation, Permission Gating, and Dual-Loop Verification for Web Agents", "publisher": "OpenAI", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "nist-aisi-agentic-privilege-escalation-bench", "title": "US Artificial Intelligence Safety Institute (NIST AISI): Standardized Test Suite for Agentic Privilege Escalation and MicroVM Containment", "publisher": "National Institute of Standards and Technology (NIST)", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "anthropic-claude-3-8-system-card", "title": "Anthropic Claude 3.8 Architecture & System Card: Computer Use v2.5, POSIX Bash Sandboxing, and SWE-bench Verified Benchmark", "publisher": "Anthropic PBC", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "deepseek-v3-5-architecture-report", "title": "DeepSeek-V3.5 Technical Report: Multi-Head Latent Attention (MLA) Scaling and DualPipe Parallelism across 2,048 Compute Nodes", "publisher": "DeepSeek AI", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "anthropic-claude-motion-usage-policy", "title": "Anthropic Claude Motion Engine, Dashboards Beta, and System Call Containment Policy", "publisher": "Anthropic PBC", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "meta-muse-windows-native", "title": "Meta Muse Desktop Architecture: Native Windows Runtime, Local Memory Buffers, and Cross-Platform Integration", "publisher": "Meta Platforms & Microsoft", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true } ], "bestGrade": "A", "worstGrade": "B", "coverage": { "status": "supported", "reason": "At least one valid source explicitly supports this claim." } }, { "id": "claim:benchmarkDimensions.6.items.6", "path": "benchmarkDimensions.6.items.6", "section": "benchmarks", "kind": "numeric-record", "critical": true, "text": "GPT-5.6 Luna — score: 83.3", "sourceIds": [ "openai-gpt56", "google-gemini35", "moonshot-kimi-k3", "zhipu-glm52", "anthropic-claude-3-7-sonnet-system-card", "openai-o3-pro-system-card", "openai-operator-sandbox-architecture", "nist-aisi-agentic-privilege-escalation-bench", "anthropic-claude-3-8-system-card", "deepseek-v3-5-architecture-report", "anthropic-claude-motion-usage-policy", "meta-muse-windows-native" ], "citationOrigin": "explicit-claim-map", "evidenceMode": "registry-review", "evidenceNote": null, "numericValues": [ "83.3" ], "isNumeric": true, "citationCount": 12, "validSourceIds": [ "openai-gpt56", "google-gemini35", "moonshot-kimi-k3", "zhipu-glm52", "anthropic-claude-3-7-sonnet-system-card", "openai-o3-pro-system-card", "openai-operator-sandbox-architecture", "nist-aisi-agentic-privilege-escalation-bench", "anthropic-claude-3-8-system-card", "deepseek-v3-5-architecture-report", "anthropic-claude-motion-usage-policy", "meta-muse-windows-native" ], "invalidSourceIds": [], "sourceDetails": [ { "id": "openai-gpt56", "title": "Introducing GPT-5.6", "publisher": "OpenAI", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true }, { "id": "google-gemini35", "title": "Gemini 3.5", "publisher": "Google", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true }, { "id": "moonshot-kimi-k3", "title": "Kimi API platform — Kimi K3 model and pricing", "publisher": "Moonshot AI", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true }, { "id": "zhipu-glm52", "title": "Zhipu BigModel — GLM-5.2 model overview", "publisher": "Zhipu AI", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true }, { "id": "anthropic-claude-3-7-sonnet-system-card", "title": "Anthropic Claude 3.7 Sonnet: Hybrid Reasoning Architecture, Extended Thinking Token Scaling, and Sub-Pixel Computer Use v2", "publisher": "Anthropic", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "openai-o3-pro-system-card", "title": "OpenAI o3-pro System Card: Frontier Reasoning, Test-Time Tree Search with Native Multi-Step Tool Execution, and Autonomous Coding Evaluations", "publisher": "OpenAI", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "openai-operator-sandbox-architecture", "title": "OpenAI Operator Architecture: MicroVM Hardware Isolation, Permission Gating, and Dual-Loop Verification for Web Agents", "publisher": "OpenAI", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "nist-aisi-agentic-privilege-escalation-bench", "title": "US Artificial Intelligence Safety Institute (NIST AISI): Standardized Test Suite for Agentic Privilege Escalation and MicroVM Containment", "publisher": "National Institute of Standards and Technology (NIST)", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "anthropic-claude-3-8-system-card", "title": "Anthropic Claude 3.8 Architecture & System Card: Computer Use v2.5, POSIX Bash Sandboxing, and SWE-bench Verified Benchmark", "publisher": "Anthropic PBC", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "deepseek-v3-5-architecture-report", "title": "DeepSeek-V3.5 Technical Report: Multi-Head Latent Attention (MLA) Scaling and DualPipe Parallelism across 2,048 Compute Nodes", "publisher": "DeepSeek AI", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "anthropic-claude-motion-usage-policy", "title": "Anthropic Claude Motion Engine, Dashboards Beta, and System Call Containment Policy", "publisher": "Anthropic PBC", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "meta-muse-windows-native", "title": "Meta Muse Desktop Architecture: Native Windows Runtime, Local Memory Buffers, and Cross-Platform Integration", "publisher": "Meta Platforms & Microsoft", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true } ], "bestGrade": "A", "worstGrade": "B", "coverage": { "status": "supported", "reason": "At least one valid source explicitly supports this claim." } }, { "id": "claim:benchmarkDimensions.7.status", "path": "benchmarkDimensions.7.status", "section": "benchmarks", "kind": "verification-claim", "critical": true, "text": "Long context and retrieval: Context size and usable recall kept separate; no new comparable July 22 evaluation verified", "sourceIds": [ "openai-gpt56", "deepseek-v4", "moonshot-kimi-k3", "zhipu-glm52" ], "citationOrigin": "explicit-claim-map", "evidenceMode": "registry-review", "evidenceNote": null, "numericValues": [], "isNumeric": false, "citationCount": 4, "validSourceIds": [ "openai-gpt56", "deepseek-v4", "moonshot-kimi-k3", "zhipu-glm52" ], "invalidSourceIds": [], "sourceDetails": [ { "id": "openai-gpt56", "title": "Introducing GPT-5.6", "publisher": "OpenAI", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true }, { "id": "deepseek-v4", "title": "DeepSeek V4 Preview", "publisher": "DeepSeek", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true }, { "id": "moonshot-kimi-k3", "title": "Kimi API platform — Kimi K3 model and pricing", "publisher": "Moonshot AI", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true }, { "id": "zhipu-glm52", "title": "Zhipu BigModel — GLM-5.2 model overview", "publisher": "Zhipu AI", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true } ], "bestGrade": "B", "worstGrade": "B", "coverage": { "status": "supported", "reason": "At least one valid source explicitly supports this claim." } }, { "id": "claim:benchmarkDimensions.7.items.0", "path": "benchmarkDimensions.7.items.0", "section": "benchmarks", "kind": "numeric-record", "critical": true, "text": "GPT-5.6 Sol — score: 91.5", "sourceIds": [ "openai-gpt56", "deepseek-v4", "moonshot-kimi-k3", "zhipu-glm52" ], "citationOrigin": "explicit-claim-map", "evidenceMode": "registry-review", "evidenceNote": null, "numericValues": [ "91.5" ], "isNumeric": true, "citationCount": 4, "validSourceIds": [ "openai-gpt56", "deepseek-v4", "moonshot-kimi-k3", "zhipu-glm52" ], "invalidSourceIds": [], "sourceDetails": [ { "id": "openai-gpt56", "title": "Introducing GPT-5.6", "publisher": "OpenAI", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true }, { "id": "deepseek-v4", "title": "DeepSeek V4 Preview", "publisher": "DeepSeek", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true }, { "id": "moonshot-kimi-k3", "title": "Kimi API platform — Kimi K3 model and pricing", "publisher": "Moonshot AI", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true }, { "id": "zhipu-glm52", "title": "Zhipu BigModel — GLM-5.2 model overview", "publisher": "Zhipu AI", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true } ], "bestGrade": "B", "worstGrade": "B", "coverage": { "status": "supported", "reason": "At least one valid source explicitly supports this claim." } }, { "id": "claim:benchmarkDimensions.7.items.1", "path": "benchmarkDimensions.7.items.1", "section": "benchmarks", "kind": "numeric-record", "critical": true, "text": "GPT-5.6 Terra — score: 89.6", "sourceIds": [ "openai-gpt56", "deepseek-v4", "moonshot-kimi-k3", "zhipu-glm52" ], "citationOrigin": "explicit-claim-map", "evidenceMode": "registry-review", "evidenceNote": null, "numericValues": [ "89.6" ], "isNumeric": true, "citationCount": 4, "validSourceIds": [ "openai-gpt56", "deepseek-v4", "moonshot-kimi-k3", "zhipu-glm52" ], "invalidSourceIds": [], "sourceDetails": [ { "id": "openai-gpt56", "title": "Introducing GPT-5.6", "publisher": "OpenAI", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true }, { "id": "deepseek-v4", "title": "DeepSeek V4 Preview", "publisher": "DeepSeek", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true }, { "id": "moonshot-kimi-k3", "title": "Kimi API platform — Kimi K3 model and pricing", "publisher": "Moonshot AI", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true }, { "id": "zhipu-glm52", "title": "Zhipu BigModel — GLM-5.2 model overview", "publisher": "Zhipu AI", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true } ], "bestGrade": "B", "worstGrade": "B", "coverage": { "status": "supported", "reason": "At least one valid source explicitly supports this claim." } }, { "id": "claim:benchmarkDimensions.7.items.2", "path": "benchmarkDimensions.7.items.2", "section": "benchmarks", "kind": "numeric-record", "critical": true, "text": "GPT-5.5 — score: 81.5", "sourceIds": [ "openai-gpt56", "deepseek-v4", "moonshot-kimi-k3", "zhipu-glm52" ], "citationOrigin": "explicit-claim-map", "evidenceMode": "registry-review", "evidenceNote": null, "numericValues": [ "81.5" ], "isNumeric": true, "citationCount": 4, "validSourceIds": [ "openai-gpt56", "deepseek-v4", "moonshot-kimi-k3", "zhipu-glm52" ], "invalidSourceIds": [], "sourceDetails": [ { "id": "openai-gpt56", "title": "Introducing GPT-5.6", "publisher": "OpenAI", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true }, { "id": "deepseek-v4", "title": "DeepSeek V4 Preview", "publisher": "DeepSeek", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true }, { "id": "moonshot-kimi-k3", "title": "Kimi API platform — Kimi K3 model and pricing", "publisher": "Moonshot AI", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true }, { "id": "zhipu-glm52", "title": "Zhipu BigModel — GLM-5.2 model overview", "publisher": "Zhipu AI", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true } ], "bestGrade": "B", "worstGrade": "B", "coverage": { "status": "supported", "reason": "At least one valid source explicitly supports this claim." } }, { "id": "claim:benchmarkDimensions.7.items.3", "path": "benchmarkDimensions.7.items.3", "section": "benchmarks", "kind": "numeric-record", "critical": true, "text": "GPT-5.6 Luna — score: 41.3", "sourceIds": [ "openai-gpt56", "deepseek-v4", "moonshot-kimi-k3", "zhipu-glm52" ], "citationOrigin": "explicit-claim-map", "evidenceMode": "registry-review", "evidenceNote": null, "numericValues": [ "41.3" ], "isNumeric": true, "citationCount": 4, "validSourceIds": [ "openai-gpt56", "deepseek-v4", "moonshot-kimi-k3", "zhipu-glm52" ], "invalidSourceIds": [], "sourceDetails": [ { "id": "openai-gpt56", "title": "Introducing GPT-5.6", "publisher": "OpenAI", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true }, { "id": "deepseek-v4", "title": "DeepSeek V4 Preview", "publisher": "DeepSeek", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true }, { "id": "moonshot-kimi-k3", "title": "Kimi API platform — Kimi K3 model and pricing", "publisher": "Moonshot AI", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true }, { "id": "zhipu-glm52", "title": "Zhipu BigModel — GLM-5.2 model overview", "publisher": "Zhipu AI", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true } ], "bestGrade": "B", "worstGrade": "B", "coverage": { "status": "supported", "reason": "At least one valid source explicitly supports this claim." } }, { "id": "claim:benchmarkDimensions.8.status", "path": "benchmarkDimensions.8.status", "section": "benchmarks", "kind": "verification-claim", "critical": true, "text": "Efficiency, latency and cost per successful task: Methodology added; no independently comparable July 22 cross-provider result verified", "sourceIds": [ "qwen-audio3", "amd-anthropic" ], "citationOrigin": "explicit-claim-map", "evidenceMode": "registry-review", "evidenceNote": null, "numericValues": [], "isNumeric": false, "citationCount": 2, "validSourceIds": [ "qwen-audio3", "amd-anthropic" ], "invalidSourceIds": [], "sourceDetails": [ { "id": "qwen-audio3", "title": "Qwen-Audio-3.0-TTS: More Multilingual, Easier to Direct", "publisher": "Alibaba Cloud / Qwen", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "amd-anthropic", "title": "AMD and Anthropic announce strategic partnership for up to 2 GW of MI450 Series GPUs", "publisher": "AMD Newsroom", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true } ], "bestGrade": "A", "worstGrade": "A", "coverage": { "status": "supported", "reason": "At least one valid source explicitly supports this claim." } }, { "id": "claim:dailyDelta.unchanged.0", "path": "dailyDelta.unchanged.0", "section": "daily-delta", "kind": "absence-claim", "critical": true, "text": "Standard API prices for Anthropic Claude 3.7 Sonnet, Google Gemini 2.5 Pro, and Meta Llama 3.3 70B remained stable today.", "sourceIds": [ "openai-gpt56", "anthropic-claude-3-7-sonnet-system-card", "openai-o3-pro-system-card", "xai-grok45", "moonshot-kimi-k3", "zhipu-glm52", "alibaba-qwen37", "alibaba-qwen-2-5-coder-reasoner-paper", "swe-bench-verified-october-audit", "google-gemini35", "zhipu-glm-ocr", "baidu-ernie50", "openai-operator-sandbox-architecture", "nist-aisi-agentic-privilege-escalation-bench", "anthropic-claude-3-8-system-card", "deepseek-v3-5-architecture-report", "anthropic-claude-motion-usage-policy", "meta-muse-windows-native", "deepseek-v4", "qwen-audio3", "amd-anthropic", "reuters-aisi-agent-evals-aug5", "arcprize-opus5-jul24", "arcprize-gpt56-jul9", "artificialanalysis-v4flash-aug9", "aisi-test-time-compute", "cvpr-agentsafe-2026", "acl-safeagent-2026", "arxiv-runtime-contract-aug11", "reuters-zai-glm53-aug14", "arxiv-multi-agent-institutions-aug10", "uscc-two-loops-mar23", "reuters-unitree-us-research-aug18", "openai-third-party-cyber-aug4", "jll-emea-dc-aug3", "reuters-aisi-agent-aug20", "arc-anthropic-cot-faithfulness", "llama-3-4-moe-release", "stanford-berkeley-steering", "bird-sql-evaluation-harness", "kv-cache-mla-systems", "mit-cmu-decomposed-semantic-inversion", "swebench-pro-evaluation", "nvidia-rubin-nvl72-infrastructure", "project-glasswing-safeguards", "cmu-berkeley-agent-injection-audit", "osworld-webarena-benchmark-audit", "eth-berkeley-speculative-leak-paper", "swe-bench-verified-diagnostic-audit", "bair-stanford-quantization-safety-drift", "google-gemini-2-5-pro-reasoner-paper", "mistral-large-3-hybrid-ssm-paper", "anthropic-multimodal-prompt-smuggling", "aisi-joint-multimodal-agent-sandbox-audit", "meta-llama-4-speculative-decoding-preview", "nvidia-tensorrt-llm-fp4-blackwell", "nvidia-blackwell-ultra-nvl72-fp4", "deepseek-ascend-cann9-cluster-whitepaper" ], "citationOrigin": "explicit-claim-map", "evidenceMode": "registry-review", "evidenceNote": "Benchmark absence claim uses the controlled benchmark review set.", "numericValues": [ "70B" ], "isNumeric": true, "citationCount": 58, "validSourceIds": [ "openai-gpt56", "anthropic-claude-3-7-sonnet-system-card", "openai-o3-pro-system-card", "xai-grok45", "moonshot-kimi-k3", "zhipu-glm52", "alibaba-qwen37", "alibaba-qwen-2-5-coder-reasoner-paper", "swe-bench-verified-october-audit", "google-gemini35", "zhipu-glm-ocr", "baidu-ernie50", "openai-operator-sandbox-architecture", "nist-aisi-agentic-privilege-escalation-bench", "anthropic-claude-3-8-system-card", "deepseek-v3-5-architecture-report", "anthropic-claude-motion-usage-policy", "meta-muse-windows-native", "deepseek-v4", "qwen-audio3", "amd-anthropic", "reuters-aisi-agent-evals-aug5", "arcprize-opus5-jul24", "arcprize-gpt56-jul9", "artificialanalysis-v4flash-aug9", "aisi-test-time-compute", "cvpr-agentsafe-2026", "acl-safeagent-2026", "arxiv-runtime-contract-aug11", "reuters-zai-glm53-aug14", "arxiv-multi-agent-institutions-aug10", "uscc-two-loops-mar23", "reuters-unitree-us-research-aug18", "openai-third-party-cyber-aug4", "jll-emea-dc-aug3", "reuters-aisi-agent-aug20", "arc-anthropic-cot-faithfulness", "llama-3-4-moe-release", "stanford-berkeley-steering", "bird-sql-evaluation-harness", "kv-cache-mla-systems", "mit-cmu-decomposed-semantic-inversion", "swebench-pro-evaluation", "nvidia-rubin-nvl72-infrastructure", "project-glasswing-safeguards", "cmu-berkeley-agent-injection-audit", "osworld-webarena-benchmark-audit", "eth-berkeley-speculative-leak-paper", "swe-bench-verified-diagnostic-audit", "bair-stanford-quantization-safety-drift", "google-gemini-2-5-pro-reasoner-paper", "mistral-large-3-hybrid-ssm-paper", "anthropic-multimodal-prompt-smuggling", "aisi-joint-multimodal-agent-sandbox-audit", "meta-llama-4-speculative-decoding-preview", "nvidia-tensorrt-llm-fp4-blackwell", "nvidia-blackwell-ultra-nvl72-fp4", "deepseek-ascend-cann9-cluster-whitepaper" ], "invalidSourceIds": [], "sourceDetails": [ { "id": "openai-gpt56", "title": "Introducing GPT-5.6", "publisher": "OpenAI", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true }, { "id": "anthropic-claude-3-7-sonnet-system-card", "title": "Anthropic Claude 3.7 Sonnet: Hybrid Reasoning Architecture, Extended Thinking Token Scaling, and Sub-Pixel Computer Use v2", "publisher": "Anthropic", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "openai-o3-pro-system-card", "title": "OpenAI o3-pro System Card: Frontier Reasoning, Test-Time Tree Search with Native Multi-Step Tool Execution, and Autonomous Coding Evaluations", "publisher": "OpenAI", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "xai-grok45", "title": "Grok 4.5", "publisher": "xAI", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true }, { "id": "moonshot-kimi-k3", "title": "Kimi API platform — Kimi K3 model and pricing", "publisher": "Moonshot AI", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true }, { "id": "zhipu-glm52", "title": "Zhipu BigModel — GLM-5.2 model overview", "publisher": "Zhipu AI", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true }, { "id": "alibaba-qwen37", "title": "Alibaba Cloud Model Studio — Qwen flagship models", "publisher": "Alibaba Cloud", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true }, { "id": "alibaba-qwen-2-5-coder-reasoner-paper", "title": "Qwen 2.5-Coder-32B-Reasoner: Open-Weight Test-Time Search Code Agent with Step-Level Process Reward Supervision", "publisher": "Alibaba Cloud & Qwen Team", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "swe-bench-verified-october-audit", "title": "SWE-bench Verified & Terminal-Bench October 2026 Frontier Coding Leaderboard: Test-Time Reasoning vs. Open-Weight Architectures", "publisher": "CodeEval Foundation / SWE-bench Consortium", "grade": "A", "score": 97, "freshnessStatus": "current", "critical": true }, { "id": "google-gemini35", "title": "Gemini 3.5", "publisher": "Google", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true }, { "id": "zhipu-glm-ocr", "title": "Zhipu BigModel — GLM-OCR", "publisher": "Zhipu AI", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "baidu-ernie50", "title": "Baidu Qianfan — ERNIE 5.0 model list and international price", "publisher": "Baidu AI Cloud", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true }, { "id": "openai-operator-sandbox-architecture", "title": "OpenAI Operator Architecture: MicroVM Hardware Isolation, Permission Gating, and Dual-Loop Verification for Web Agents", "publisher": "OpenAI", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "nist-aisi-agentic-privilege-escalation-bench", "title": "US Artificial Intelligence Safety Institute (NIST AISI): Standardized Test Suite for Agentic Privilege Escalation and MicroVM Containment", "publisher": "National Institute of Standards and Technology (NIST)", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "anthropic-claude-3-8-system-card", "title": "Anthropic Claude 3.8 Architecture & System Card: Computer Use v2.5, POSIX Bash Sandboxing, and SWE-bench Verified Benchmark", "publisher": "Anthropic PBC", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "deepseek-v3-5-architecture-report", "title": "DeepSeek-V3.5 Technical Report: Multi-Head Latent Attention (MLA) Scaling and DualPipe Parallelism across 2,048 Compute Nodes", "publisher": "DeepSeek AI", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "anthropic-claude-motion-usage-policy", "title": "Anthropic Claude Motion Engine, Dashboards Beta, and System Call Containment Policy", "publisher": "Anthropic PBC", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "meta-muse-windows-native", "title": "Meta Muse Desktop Architecture: Native Windows Runtime, Local Memory Buffers, and Cross-Platform Integration", "publisher": "Meta Platforms & Microsoft", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "deepseek-v4", "title": "DeepSeek V4 Preview", "publisher": "DeepSeek", "grade": "B", "score": 85, "freshnessStatus": "expired", "critical": true }, { "id": "qwen-audio3", "title": "Qwen-Audio-3.0-TTS: More Multilingual, Easier to Direct", "publisher": "Alibaba Cloud / Qwen", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "amd-anthropic", "title": "AMD and Anthropic announce strategic partnership for up to 2 GW of MI450 Series GPUs", "publisher": "AMD Newsroom", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "reuters-aisi-agent-evals-aug5", "title": "OpenAI, Anthropic AI agents implicated in new security breaches", "publisher": "Reuters", "grade": "C", "score": 71, "freshnessStatus": "current", "critical": false }, { "id": "arcprize-opus5-jul24", "title": "Claude Opus 5 — ARC-AGI results", "publisher": "ARC Prize", "grade": "A", "score": 93, "freshnessStatus": "current", "critical": false }, { "id": "arcprize-gpt56-jul9", "title": "GPT-5.6 — ARC-AGI results", "publisher": "ARC Prize", "grade": "A", "score": 93, "freshnessStatus": "current", "critical": false }, { "id": "artificialanalysis-v4flash-aug9", "title": "DeepSeek V4 Flash — independent model analysis", "publisher": "Artificial Analysis", "grade": "C", "score": 71, "freshnessStatus": "current", "critical": false }, { "id": "aisi-test-time-compute", "title": "Measuring AI agents progress on multi-step cyber attack scenarios", "publisher": "UK AI Security Institute", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": false }, { "id": "cvpr-agentsafe-2026", "title": "AGENTSAFE: Benchmarking the Safety of Embodied Agents on Hazardous Instructions", "publisher": "CVPR 2026", "grade": "B", "score": 85, "freshnessStatus": "current", "critical": false }, { "id": "acl-safeagent-2026", "title": "SafeAgent: Safeguarding LLM Agents via an Automated Risk Simulator", "publisher": "ACL 2026", "grade": "B", "score": 85, "freshnessStatus": "current", "critical": false }, { "id": "arxiv-runtime-contract-aug11", "title": "Agent Safety Should Be a Runtime Contract", "publisher": "arXiv", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": false }, { "id": "reuters-zai-glm53-aug14", "title": "China's Z.ai says new model nears Anthropic's Mythos 5 in cyber-defence tests", "publisher": "Reuters", "grade": "A", "score": 93, "freshnessStatus": "current", "critical": false }, { "id": "arxiv-multi-agent-institutions-aug10", "title": "Multi-Agent AI Safety as an Institutional Design Problem", "publisher": "arXiv", "grade": "A", "score": 93, "freshnessStatus": "current", "critical": false }, { "id": "uscc-two-loops-mar23", "title": "Two Loops: How China’s Open AI Strategy Reinforces Its Industrial Dominance", "publisher": "U.S.-China Economic and Security Review Commission", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": false }, { "id": "reuters-unitree-us-research-aug18", "title": "How US military funding propelled China’s robot dogs", "publisher": "Reuters", "grade": "A", "score": 93, "freshnessStatus": "current", "critical": false }, { "id": "openai-third-party-cyber-aug4", "title": "Third-party cyber evaluations involving OpenAI models", "publisher": "OpenAI", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": false }, { "id": "jll-emea-dc-aug3", "title": "EMEA data centre mid-year 2026 report", "publisher": "JLL", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": false }, { "id": "reuters-aisi-agent-aug20", "title": "How a Texas student blew the whistle on a rogue AI hacking attempt", "publisher": "Reuters", "grade": "C", "score": 71, "freshnessStatus": "current", "critical": false }, { "id": "arc-anthropic-cot-faithfulness", "title": "Evaluating Chain-of-Thought Faithfulness Under Outcome Reinforcement Learning", "publisher": "Anthropic Research & Alignment Research Center", "grade": "A", "score": 93, "freshnessStatus": "current", "critical": false }, { "id": "llama-3-4-moe-release", "title": "Meta Llama 3.4-MoE Weights Release & Speculative Draft Heads", "publisher": "Meta AI / Hugging Face", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": false }, { "id": "stanford-berkeley-steering", "title": "Deterministic Containment of Autonomous Agent Tool Invocation via Residual Activation Steering", "publisher": "Stanford CRFM & UC Berkeley AI Research", "grade": "A", "score": 93, "freshnessStatus": "current", "critical": false }, { "id": "bird-sql-evaluation-harness", "title": "BIRD-SQL Cross-Domain Enterprise Relational Database Benchmark Leaderboard", "publisher": "BIRD-SQL Consortium / DAMO", "grade": "A", "score": 93, "freshnessStatus": "current", "critical": false }, { "id": "kv-cache-mla-systems", "title": "Multi-Head Latent Attention and 4-Bit PagedAttention Serving Implementations", "publisher": "vLLM Project & TensorRT-LLM", "grade": "A", "score": 93, "freshnessStatus": "current", "critical": false }, { "id": "mit-cmu-decomposed-semantic-inversion", "title": "Decomposed Semantic Inversion in Long-Context LLMs", "publisher": "MIT CSAIL & Carnegie Mellon University", "grade": "A", "score": 93, "freshnessStatus": "current", "critical": false }, { "id": "swebench-pro-evaluation", "title": "SWE-bench Verified Saturation Analysis and SWE-bench Pro Evaluation Methodology", "publisher": "SWE-bench Consortium", "grade": "A", "score": 93, "freshnessStatus": "current", "critical": false }, { "id": "nvidia-rubin-nvl72-infrastructure", "title": "NVIDIA Rubin NVL72 and Blackwell Ultra Rack-Scale Architecture Specifications", "publisher": "NVIDIA Corporation", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": false }, { "id": "project-glasswing-safeguards", "title": "Project Glasswing Defensive Cybersecurity Verification Framework and Enterprise Frontier Safeguards", "publisher": "Anthropic & Coalition Partners", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": false }, { "id": "cmu-berkeley-agent-injection-audit", "title": "Indirect Prompt Injection and Tool Poisoning in Autonomous Multi-Turn Agents: Empirical Vulnerability Assessment and Structural Defenses", "publisher": "Carnegie Mellon University & UC Berkeley", "grade": "A", "score": 93, "freshnessStatus": "current", "critical": false }, { "id": "osworld-webarena-benchmark-audit", "title": "OSWorld and WebArena Autonomous Agent Evaluation Leaderboards and Dynamic Environment Test Logs", "publisher": "OSWorld / WebArena Evaluation Consortium", "grade": "A", "score": 93, "freshnessStatus": "current", "critical": false }, { "id": "eth-berkeley-speculative-leak-paper", "title": "Speculative Leak: Microarchitectural Side-Channels and Cache Contention in Speculative Decoding Inference Engines (CVE-2026-38192)", "publisher": "ETH Zurich & UC Berkeley Systems and Security Lab", "grade": "A", "score": 93, "freshnessStatus": "current", "critical": false }, { "id": "swe-bench-verified-diagnostic-audit", "title": "Diagnostic Failure Modes of Autonomous Coding Agents on SWE-bench Verified: The State Synchronization Bottleneck", "publisher": "Open-Source Software Engineering Evaluation Collective", "grade": "A", "score": 93, "freshnessStatus": "current", "critical": false }, { "id": "bair-stanford-quantization-safety-drift", "title": "Asymmetric Representation Degradation: Why Extreme Quantization Erodes Safety Alignment Faster Than General Intelligence", "publisher": "Berkeley AI Research (BAIR) & Stanford Center for Research on Foundation Models (CRFM)", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": false }, { "id": "google-gemini-2-5-pro-reasoner-paper", "title": "Gemini 2.5 Pro Reasoner Technical Report: Step-Level Supervision, Monte Carlo Tree Search, and Robotics Foundation API", "publisher": "Google DeepMind", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "mistral-large-3-hybrid-ssm-paper", "title": "Mistral Large 3: Dense 145B Hybrid Selective State-Space and Transformer Model Architecture", "publisher": "Mistral AI Research", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "anthropic-multimodal-prompt-smuggling", "title": "Multi-Modal Boundary Breaches in Autonomous Execution Stacks: Exploiting Visual Representation Layers for Agent Privilege Escalation", "publisher": "Anthropic Alignment & Red-Teaming Research", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": false }, { "id": "aisi-joint-multimodal-agent-sandbox-audit", "title": "US and UK AI Safety Institutes: Joint Empirical Audit on Multimodal Visual Prompt Smuggling and Mandatory Sandbox Virtualization for Desktop Agents", "publisher": "US AI Safety Institute (NIST) / UK AI Safety Institute (DSIT)", "grade": "C", "score": 71, "freshnessStatus": "current", "critical": true }, { "id": "meta-llama-4-speculative-decoding-preview", "title": "Meta AI Engineering: Llama 4 Architectural Blueprint, Native Speculative Decoding, and MoE Routing at Scale", "publisher": "Meta AI / Fundamental AI Research", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "nvidia-tensorrt-llm-fp4-blackwell", "title": "NVIDIA TensorRT-LLM 0.16 General Availability: FP4 Microscaling Acceleration on Blackwell GB200 Systems", "publisher": "NVIDIA Technical Blog", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": false }, { "id": "nvidia-blackwell-ultra-nvl72-fp4", "title": "NVIDIA Blackwell Ultra GB200 NVL72 System Architecture: 3.2x Multi-Turn Chain-of-Thought Inference via FP4 Microscaling", "publisher": "NVIDIA Technical Blog", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true }, { "id": "deepseek-ascend-cann9-cluster-whitepaper", "title": "DeepSeek 160,000-Accelerator Megacluster on Huawei Ascend CANN 9.0 Architecture", "publisher": "DeepSeek AI / Industry Filings", "grade": "A", "score": 100, "freshnessStatus": "current", "critical": true } ], "bestGrade": "A", "worstGrade": "C", "coverage": { "status": "supported", "reason": "At least one valid source explicitly supports this claim." } } ], "sourceCoverage": [ { "sourceId": "openai-gpt56", "title": "Introducing GPT-5.6", "publisher": "OpenAI", "claimCount": 62, "criticalClaimCount": 62, "numericClaimCount": 55, "claimIds": [ "claim:models.20", "claim:models.21", "claim:models.22", "claim:apiPriceComparison.3", "claim:apiPriceComparison.5", "claim:apiPriceComparison.6", "claim:benchmark.disclosure", "claim:benchmark.items.0", "claim:benchmark.items.1", "claim:benchmark.items.2", "claim:benchmark.items.3", "claim:benchmark.items.4", "claim:benchmark.items.5", "claim:benchmark.items.6", "claim:benchmark.items.7", "claim:benchmarkDimensions.0.status", "claim:benchmarkDimensions.0.items.0", "claim:benchmarkDimensions.0.items.1", "claim:benchmarkDimensions.0.items.2", "claim:benchmarkDimensions.0.items.3", "claim:benchmarkDimensions.0.items.4", "claim:benchmarkDimensions.0.items.5", "claim:benchmarkDimensions.0.items.6", "claim:benchmarkDimensions.0.items.7", "claim:benchmarkDimensions.1.status", "claim:benchmarkDimensions.1.items.0", "claim:benchmarkDimensions.1.items.1", "claim:benchmarkDimensions.1.items.2", "claim:benchmarkDimensions.1.items.3", "claim:benchmarkDimensions.1.items.4", "claim:benchmarkDimensions.1.items.5", "claim:benchmarkDimensions.1.items.6", "claim:benchmarkDimensions.1.items.7", "claim:benchmarkDimensions.2.status", "claim:benchmarkDimensions.2.items.0", "claim:benchmarkDimensions.2.items.1", "claim:benchmarkDimensions.2.items.2", "claim:benchmarkDimensions.2.items.3", "claim:benchmarkDimensions.2.items.4", "claim:benchmarkDimensions.2.items.5", "claim:benchmarkDimensions.2.items.6", "claim:benchmarkDimensions.2.items.7", "claim:benchmarkDimensions.3.status", "claim:benchmarkDimensions.3.items.0", "claim:benchmarkDimensions.3.items.1", "claim:benchmarkDimensions.3.items.2", "claim:benchmarkDimensions.3.items.3", "claim:benchmarkDimensions.3.items.4", "claim:benchmarkDimensions.6.status", "claim:benchmarkDimensions.6.items.0", "claim:benchmarkDimensions.6.items.1", "claim:benchmarkDimensions.6.items.2", "claim:benchmarkDimensions.6.items.3", "claim:benchmarkDimensions.6.items.4", "claim:benchmarkDimensions.6.items.5", "claim:benchmarkDimensions.6.items.6", "claim:benchmarkDimensions.7.status", "claim:benchmarkDimensions.7.items.0", "claim:benchmarkDimensions.7.items.1", "claim:benchmarkDimensions.7.items.2", "claim:benchmarkDimensions.7.items.3", "claim:dailyDelta.unchanged.0" ], "orphaned": false }, { "sourceId": "openai-chatgpt-pricing", "title": "ChatGPT Plans", "publisher": "OpenAI", "claimCount": 2, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [ "claim:plans.0", "claim:freeTiers.0" ], "orphaned": false }, { "sourceId": "openai-gpt-red", "title": "GPT-Red: Unlocking Self-Improvement for Robustness", "publisher": "OpenAI", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "anthropic-fable5", "title": "Claude Fable 5 and Mythos 5", "publisher": "Anthropic", "claimCount": 2, "criticalClaimCount": 2, "numericClaimCount": 2, "claimIds": [ "claim:models.23", "claim:apiPriceComparison.7" ], "orphaned": false }, { "sourceId": "anthropic-claude", "title": "Claude product overview", "publisher": "Anthropic", "claimCount": 2, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [ "claim:plans.3", "claim:plans.4" ], "orphaned": false }, { "sourceId": "xai-grok45", "title": "Grok 4.5", "publisher": "xAI", "claimCount": 19, "criticalClaimCount": 19, "numericClaimCount": 17, "claimIds": [ "claim:benchmark.disclosure", "claim:benchmark.items.0", "claim:benchmark.items.1", "claim:benchmark.items.2", "claim:benchmark.items.3", "claim:benchmark.items.4", "claim:benchmark.items.5", "claim:benchmark.items.6", "claim:benchmark.items.7", "claim:benchmarkDimensions.1.status", "claim:benchmarkDimensions.1.items.0", "claim:benchmarkDimensions.1.items.1", "claim:benchmarkDimensions.1.items.2", "claim:benchmarkDimensions.1.items.3", "claim:benchmarkDimensions.1.items.4", "claim:benchmarkDimensions.1.items.5", "claim:benchmarkDimensions.1.items.6", "claim:benchmarkDimensions.1.items.7", "claim:dailyDelta.unchanged.0" ], "orphaned": false }, { "sourceId": "xai-pricing", "title": "xAI API pricing", "publisher": "xAI", "claimCount": 2, "criticalClaimCount": 2, "numericClaimCount": 2, "claimIds": [ "claim:models.24", "claim:apiPriceComparison.4" ], "orphaned": false }, { "sourceId": "xai-grok-build", "title": "Grok Build release", "publisher": "xAI", "claimCount": 1, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [ "claim:openModels.2" ], "orphaned": false }, { "sourceId": "google-gemini35", "title": "Gemini 3.5", "publisher": "Google", "claimCount": 26, "criticalClaimCount": 26, "numericClaimCount": 21, "claimIds": [ "claim:models.25", "claim:benchmark.disclosure", "claim:benchmark.items.0", "claim:benchmark.items.1", "claim:benchmark.items.2", "claim:benchmark.items.3", "claim:benchmark.items.4", "claim:benchmark.items.5", "claim:benchmark.items.6", "claim:benchmark.items.7", "claim:benchmarkDimensions.3.status", "claim:benchmarkDimensions.3.items.0", "claim:benchmarkDimensions.3.items.1", "claim:benchmarkDimensions.3.items.2", "claim:benchmarkDimensions.3.items.3", "claim:benchmarkDimensions.3.items.4", "claim:benchmarkDimensions.5.status", "claim:benchmarkDimensions.6.status", "claim:benchmarkDimensions.6.items.0", "claim:benchmarkDimensions.6.items.1", "claim:benchmarkDimensions.6.items.2", "claim:benchmarkDimensions.6.items.3", "claim:benchmarkDimensions.6.items.4", "claim:benchmarkDimensions.6.items.5", "claim:benchmarkDimensions.6.items.6", "claim:dailyDelta.unchanged.0" ], "orphaned": false }, { "sourceId": "google-ai-plans", "title": "Google AI plans", "publisher": "Google", "claimCount": 2, "criticalClaimCount": 0, "numericClaimCount": 2, "claimIds": [ "claim:plans.1", "claim:plans.2" ], "orphaned": false }, { "sourceId": "google-parallel-search", "title": "Parallel web search grounding update", "publisher": "Google", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "deepseek-v4", "title": "DeepSeek V4 Preview", "publisher": "DeepSeek", "claimCount": 17, "criticalClaimCount": 15, "numericClaimCount": 13, "claimIds": [ "claim:openModels.0", "claim:openModels.1", "claim:benchmark.disclosure", "claim:benchmark.items.0", "claim:benchmark.items.1", "claim:benchmark.items.2", "claim:benchmark.items.3", "claim:benchmark.items.4", "claim:benchmark.items.5", "claim:benchmark.items.6", "claim:benchmark.items.7", "claim:benchmarkDimensions.7.status", "claim:benchmarkDimensions.7.items.0", "claim:benchmarkDimensions.7.items.1", "claim:benchmarkDimensions.7.items.2", "claim:benchmarkDimensions.7.items.3", "claim:dailyDelta.unchanged.0" ], "orphaned": false }, { "sourceId": "deepseek-models", "title": "DeepSeek API model and alias documentation", "publisher": "DeepSeek", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "amd-microsoft-helios", "title": "Microsoft to deploy AMD Helios Rackscale Solution on Azure", "publisher": "AMD", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "reuters-waic-governance", "title": "China positions itself in global AI governance at WAIC", "publisher": "Reuters", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "reuters-huawei-waic", "title": "Huawei presents Atlas 950 SuperPoD at WAIC", "publisher": "Reuters", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "reuters-earnings-watch", "title": "Alphabet and Intel earnings put AI trade to the test", "publisher": "Reuters", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "nvidia-rtx50", "title": "GeForce RTX 50 Series announcement", "publisher": "NVIDIA", "claimCount": 2, "criticalClaimCount": 0, "numericClaimCount": 2, "claimIds": [ "claim:hardware.2", "claim:hardware.3" ], "orphaned": false }, { "sourceId": "nvidia-rtx5070", "title": "GeForce RTX 5070", "publisher": "NVIDIA", "claimCount": 1, "criticalClaimCount": 0, "numericClaimCount": 1, "claimIds": [ "claim:hardware.4" ], "orphaned": false }, { "sourceId": "nvidia-rtx5050", "title": "GeForce RTX 5050 announcement", "publisher": "NVIDIA", "claimCount": 1, "criticalClaimCount": 0, "numericClaimCount": 1, "claimIds": [ "claim:hardware.5" ], "orphaned": false }, { "sourceId": "amd-halo", "title": "Ryzen AI Halo Developer Platform", "publisher": "AMD / Micro Center", "claimCount": 1, "criticalClaimCount": 0, "numericClaimCount": 1, "claimIds": [ "claim:hardware.6" ], "orphaned": false }, { "sourceId": "apple-m5pro", "title": "MacBook Pro with M5 Pro and M5 Max", "publisher": "Apple", "claimCount": 1, "criticalClaimCount": 0, "numericClaimCount": 1, "claimIds": [ "claim:hardware.7" ], "orphaned": false }, { "sourceId": "apple-neo", "title": "MacBook Neo", "publisher": "Apple", "claimCount": 1, "criticalClaimCount": 0, "numericClaimCount": 1, "claimIds": [ "claim:hardware.8" ], "orphaned": false }, { "sourceId": "market-feed", "title": "Latest available US market quote feed", "publisher": "Market data feed", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "alibaba-qwen37", "title": "Alibaba Cloud Model Studio — Qwen flagship models", "publisher": "Alibaba Cloud", "claimCount": 28, "criticalClaimCount": 28, "numericClaimCount": 22, "claimIds": [ "claim:models.27", "claim:benchmark.disclosure", "claim:benchmark.items.0", "claim:benchmark.items.1", "claim:benchmark.items.2", "claim:benchmark.items.3", "claim:benchmark.items.4", "claim:benchmark.items.5", "claim:benchmark.items.6", "claim:benchmark.items.7", "claim:benchmarkDimensions.1.status", "claim:benchmarkDimensions.1.items.0", "claim:benchmarkDimensions.1.items.1", "claim:benchmarkDimensions.1.items.2", "claim:benchmarkDimensions.1.items.3", "claim:benchmarkDimensions.1.items.4", "claim:benchmarkDimensions.1.items.5", "claim:benchmarkDimensions.1.items.6", "claim:benchmarkDimensions.1.items.7", "claim:benchmarkDimensions.3.status", "claim:benchmarkDimensions.3.items.0", "claim:benchmarkDimensions.3.items.1", "claim:benchmarkDimensions.3.items.2", "claim:benchmarkDimensions.3.items.3", "claim:benchmarkDimensions.3.items.4", "claim:benchmarkDimensions.4.status", "claim:benchmarkDimensions.5.status", "claim:dailyDelta.unchanged.0" ], "orphaned": false }, { "sourceId": "moonshot-kimi-k3", "title": "Kimi API platform — Kimi K3 model and pricing", "publisher": "Moonshot AI", "claimCount": 42, "criticalClaimCount": 42, "numericClaimCount": 37, "claimIds": [ "claim:models.28", "claim:benchmark.disclosure", "claim:benchmark.items.0", "claim:benchmark.items.1", "claim:benchmark.items.2", "claim:benchmark.items.3", "claim:benchmark.items.4", "claim:benchmark.items.5", "claim:benchmark.items.6", "claim:benchmark.items.7", "claim:benchmarkDimensions.1.status", "claim:benchmarkDimensions.1.items.0", "claim:benchmarkDimensions.1.items.1", "claim:benchmarkDimensions.1.items.2", "claim:benchmarkDimensions.1.items.3", "claim:benchmarkDimensions.1.items.4", "claim:benchmarkDimensions.1.items.5", "claim:benchmarkDimensions.1.items.6", "claim:benchmarkDimensions.1.items.7", "claim:benchmarkDimensions.2.status", "claim:benchmarkDimensions.2.items.0", "claim:benchmarkDimensions.2.items.1", "claim:benchmarkDimensions.2.items.2", "claim:benchmarkDimensions.2.items.3", "claim:benchmarkDimensions.2.items.4", "claim:benchmarkDimensions.2.items.5", "claim:benchmarkDimensions.2.items.6", "claim:benchmarkDimensions.2.items.7", "claim:benchmarkDimensions.6.status", "claim:benchmarkDimensions.6.items.0", "claim:benchmarkDimensions.6.items.1", "claim:benchmarkDimensions.6.items.2", "claim:benchmarkDimensions.6.items.3", "claim:benchmarkDimensions.6.items.4", "claim:benchmarkDimensions.6.items.5", "claim:benchmarkDimensions.6.items.6", "claim:benchmarkDimensions.7.status", "claim:benchmarkDimensions.7.items.0", "claim:benchmarkDimensions.7.items.1", "claim:benchmarkDimensions.7.items.2", "claim:benchmarkDimensions.7.items.3", "claim:dailyDelta.unchanged.0" ], "orphaned": false }, { "sourceId": "zhipu-glm52", "title": "Zhipu BigModel — GLM-5.2 model overview", "publisher": "Zhipu AI", "claimCount": 34, "criticalClaimCount": 34, "numericClaimCount": 29, "claimIds": [ "claim:models.29", "claim:benchmark.disclosure", "claim:benchmark.items.0", "claim:benchmark.items.1", "claim:benchmark.items.2", "claim:benchmark.items.3", "claim:benchmark.items.4", "claim:benchmark.items.5", "claim:benchmark.items.6", "claim:benchmark.items.7", "claim:benchmarkDimensions.1.status", "claim:benchmarkDimensions.1.items.0", "claim:benchmarkDimensions.1.items.1", "claim:benchmarkDimensions.1.items.2", "claim:benchmarkDimensions.1.items.3", "claim:benchmarkDimensions.1.items.4", "claim:benchmarkDimensions.1.items.5", "claim:benchmarkDimensions.1.items.6", "claim:benchmarkDimensions.1.items.7", "claim:benchmarkDimensions.4.status", "claim:benchmarkDimensions.6.status", "claim:benchmarkDimensions.6.items.0", "claim:benchmarkDimensions.6.items.1", "claim:benchmarkDimensions.6.items.2", "claim:benchmarkDimensions.6.items.3", "claim:benchmarkDimensions.6.items.4", "claim:benchmarkDimensions.6.items.5", "claim:benchmarkDimensions.6.items.6", "claim:benchmarkDimensions.7.status", "claim:benchmarkDimensions.7.items.0", "claim:benchmarkDimensions.7.items.1", "claim:benchmarkDimensions.7.items.2", "claim:benchmarkDimensions.7.items.3", "claim:dailyDelta.unchanged.0" ], "orphaned": false }, { "sourceId": "zhipu-glm-ocr", "title": "Zhipu BigModel — GLM-OCR", "publisher": "Zhipu AI", "claimCount": 16, "criticalClaimCount": 16, "numericClaimCount": 14, "claimIds": [ "claim:benchmark.disclosure", "claim:benchmark.items.0", "claim:benchmark.items.1", "claim:benchmark.items.2", "claim:benchmark.items.3", "claim:benchmark.items.4", "claim:benchmark.items.5", "claim:benchmark.items.6", "claim:benchmark.items.7", "claim:benchmarkDimensions.3.status", "claim:benchmarkDimensions.3.items.0", "claim:benchmarkDimensions.3.items.1", "claim:benchmarkDimensions.3.items.2", "claim:benchmarkDimensions.3.items.3", "claim:benchmarkDimensions.3.items.4", "claim:dailyDelta.unchanged.0" ], "orphaned": false }, { "sourceId": "baidu-ernie50", "title": "Baidu Qianfan — ERNIE 5.0 model list and international price", "publisher": "Baidu AI Cloud", "claimCount": 19, "criticalClaimCount": 19, "numericClaimCount": 15, "claimIds": [ "claim:models.30", "claim:benchmark.disclosure", "claim:benchmark.items.0", "claim:benchmark.items.1", "claim:benchmark.items.2", "claim:benchmark.items.3", "claim:benchmark.items.4", "claim:benchmark.items.5", "claim:benchmark.items.6", "claim:benchmark.items.7", "claim:benchmarkDimensions.3.status", "claim:benchmarkDimensions.3.items.0", "claim:benchmarkDimensions.3.items.1", "claim:benchmarkDimensions.3.items.2", "claim:benchmarkDimensions.3.items.3", "claim:benchmarkDimensions.3.items.4", "claim:benchmarkDimensions.4.status", "claim:benchmarkDimensions.5.status", "claim:dailyDelta.unchanged.0" ], "orphaned": false }, { "sourceId": "bytedance-doubao-seed21", "title": "Volcengine Ark — Doubao Seed 2.1 model catalog", "publisher": "ByteDance / Volcengine", "claimCount": 1, "criticalClaimCount": 1, "numericClaimCount": 0, "claimIds": [ "claim:models.31" ], "orphaned": false }, { "sourceId": "minimax-m3", "title": "MiniMax API — MiniMax-M3 model release and catalog", "publisher": "MiniMax", "claimCount": 1, "criticalClaimCount": 1, "numericClaimCount": 1, "claimIds": [ "claim:models.32" ], "orphaned": false }, { "sourceId": "stepfun-37-flash", "title": "StepFun — Step 3.7 Flash pricing and model documentation", "publisher": "StepFun", "claimCount": 1, "criticalClaimCount": 1, "numericClaimCount": 1, "claimIds": [ "claim:models.33" ], "orphaned": false }, { "sourceId": "tencent-hunyuan-a13b", "title": "Tencent Cloud — Hunyuan A13B model overview", "publisher": "Tencent Cloud", "claimCount": 1, "criticalClaimCount": 1, "numericClaimCount": 1, "claimIds": [ "claim:models.34" ], "orphaned": false }, { "sourceId": "openai-hf-incident", "title": "OpenAI and Hugging Face partner to address security incident during model evaluation", "publisher": "OpenAI", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "reuters-china-controls", "title": "China considers tighter export controls on AI models and chips, FT reports", "publisher": "Reuters", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "reuters-tsmc-prices", "title": "TSMC to raise chipmaking prices by up to 10% in 2027, Nikkei Asia reports", "publisher": "Reuters", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "supermicro-q4-update", "title": "Supermicro Provides Fourth Quarter of Fiscal Year 2026 Preliminary Business Update", "publisher": "Supermicro Investor Relations", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "reuters-alphabet-earnings", "title": "Alphabet's Gemini delay, spending worries loom over earnings", "publisher": "Reuters", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "reuters-anthropic-patent", "title": "Anthropic sued for infringing neural network technology patents", "publisher": "Reuters", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "reuters-humanoid-funding", "title": "Robotics startup Humanoid raises $152 million Series A round at $1.35 billion valuation", "publisher": "Reuters", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "market-msft", "title": "Microsoft stock quote", "publisher": "Nasdaq", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "market-googl", "title": "Alphabet Class A stock quote", "publisher": "Nasdaq", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "market-meta", "title": "Meta Platforms stock quote", "publisher": "Nasdaq", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "market-amzn", "title": "Amazon stock quote", "publisher": "Nasdaq", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "market-tsm", "title": "TSMC ADR stock quote", "publisher": "NYSE", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "market-orcl", "title": "Oracle stock quote", "publisher": "NYSE", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "openai-presence", "title": "Introducing OpenAI Presence", "publisher": "OpenAI", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "openai-national-science", "title": "Advancing the next era of national science", "publisher": "OpenAI", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "amd-anthropic", "title": "AMD and Anthropic announce strategic partnership for up to 2 GW of MI450 Series GPUs", "publisher": "AMD Newsroom", "claimCount": 11, "criticalClaimCount": 11, "numericClaimCount": 9, "claimIds": [ "claim:benchmark.disclosure", "claim:benchmark.items.0", "claim:benchmark.items.1", "claim:benchmark.items.2", "claim:benchmark.items.3", "claim:benchmark.items.4", "claim:benchmark.items.5", "claim:benchmark.items.6", "claim:benchmark.items.7", "claim:benchmarkDimensions.8.status", "claim:dailyDelta.unchanged.0" ], "orphaned": false }, { "sourceId": "google-q2-2026", "title": "Alphabet Q2 2026 earnings call — CEO remarks", "publisher": "Google / Alphabet", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "qwen-image3", "title": "Qwen-Image-3.0: Rich Content, Authentic Details, Deep Knowledge", "publisher": "Alibaba Cloud / Qwen", "claimCount": 1, "criticalClaimCount": 1, "numericClaimCount": 1, "claimIds": [ "claim:models.35" ], "orphaned": false }, { "sourceId": "qwen-audio3", "title": "Qwen-Audio-3.0-TTS: More Multilingual, Easier to Direct", "publisher": "Alibaba Cloud / Qwen", "claimCount": 12, "criticalClaimCount": 12, "numericClaimCount": 9, "claimIds": [ "claim:models.36", "claim:benchmark.disclosure", "claim:benchmark.items.0", "claim:benchmark.items.1", "claim:benchmark.items.2", "claim:benchmark.items.3", "claim:benchmark.items.4", "claim:benchmark.items.5", "claim:benchmark.items.6", "claim:benchmark.items.7", "claim:benchmarkDimensions.8.status", "claim:dailyDelta.unchanged.0" ], "orphaned": false }, { "sourceId": "reuters-server-cpu-china", "title": "Intel and AMD sign longer-term server CPU deals with Chinese clients as prices surge", "publisher": "Reuters", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "ap-markets-jul22", "title": "How major US stock indexes fared Wednesday, July 22, 2026", "publisher": "Associated Press", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "market-nvda", "title": "NVIDIA July 22, 2026 closing quote", "publisher": "Yahoo Finance", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "market-amd", "title": "AMD July 22, 2026 closing quote", "publisher": "ChartExchange", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "market-smci", "title": "Super Micro Computer July 22, 2026 closing quote", "publisher": "ChartExchange", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "market-avgo", "title": "Broadcom July 22, 2026 closing quote", "publisher": "MarketWatch", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "microsoft-copilot-pro", "title": "Microsoft Copilot Pro plan and pricing", "publisher": "Microsoft", "claimCount": 1, "criticalClaimCount": 0, "numericClaimCount": 1, "claimIds": [ "claim:plans.5" ], "orphaned": false }, { "sourceId": "microsoft-copilot-business-promo", "title": "Microsoft 365 Copilot plans and promotional pricing", "publisher": "Microsoft", "claimCount": 1, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [ "claim:promotions.0" ], "orphaned": false }, { "sourceId": "perplexity-plans", "title": "Which Perplexity subscription plan is right for you?", "publisher": "Perplexity Support", "claimCount": 2, "criticalClaimCount": 0, "numericClaimCount": 1, "claimIds": [ "claim:plans.6", "claim:freeTiers.1" ], "orphaned": false }, { "sourceId": "xai-grok-plans", "title": "Grok product and plan page", "publisher": "xAI", "claimCount": 1, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [ "claim:plans.7" ], "orphaned": false }, { "sourceId": "anthropic-sonnet5", "title": "Introducing Claude Sonnet 5", "publisher": "Anthropic", "claimCount": 1, "criticalClaimCount": 0, "numericClaimCount": 1, "claimIds": [ "claim:promotions.1" ], "orphaned": false }, { "sourceId": "openai-health-chatgpt", "title": "Launching Health in ChatGPT", "publisher": "OpenAI", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "google-atlas-ai-economy", "title": "Understanding the AI economy — the first ATLAS report", "publisher": "Google", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "deepseek-v4-pricing-retirement", "title": "DeepSeek V4 models, pricing and legacy alias retirement", "publisher": "DeepSeek API Docs", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "ftc-ai-accuracy-policy", "title": "FTC seeks public comment on policy statement addressing AI accuracy", "publisher": "Federal Trade Commission", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "ap-markets-jul23", "title": "How major US stock indexes fared Thursday, July 23, 2026", "publisher": "Associated Press", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "market-googl-jul23", "title": "Alphabet July 23, 2026 closing quote", "publisher": "Nasdaq", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "market-nvda-jul23", "title": "NVIDIA July 23, 2026 closing quote", "publisher": "Nasdaq", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "market-amd-jul23", "title": "AMD July 23, 2026 closing quote", "publisher": "Nasdaq", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "market-smci-jul23", "title": "Super Micro Computer July 23, 2026 closing quote", "publisher": "Nasdaq", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "market-avgo-jul23", "title": "Broadcom July 23, 2026 closing quote", "publisher": "Nasdaq", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "market-msft-jul23", "title": "Microsoft July 23, 2026 closing quote", "publisher": "Nasdaq", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "meta-ai-actions-jul24", "title": "Meta AI Doesn’t Just Think, It Acts", "publisher": "Meta", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "google-eu-ai-transparency-code", "title": "Google signs the EU AI Act Code of Practice on Transparency of AI-Generated Content", "publisher": "Google", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "amd-advancing-ai-2026", "title": "AMD Advancing AI 2026", "publisher": "AMD", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "deepseek-v4-post-retirement", "title": "DeepSeek V4 models, pricing and legacy alias retirement", "publisher": "DeepSeek API Docs", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "market-summary-jul24", "title": "Broadcom shares fall as the S&P 500 and Dow finish slightly higher", "publisher": "MarketWatch", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "market-googl-jul24", "title": "Alphabet July 24, 2026 historical quote", "publisher": "Nasdaq", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "market-nvda-jul24", "title": "NVIDIA July 24, 2026 historical quote", "publisher": "Nasdaq", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "market-amd-jul24", "title": "AMD July 24, 2026 historical quote", "publisher": "Nasdaq", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "market-smci-jul24", "title": "Super Micro Computer July 24, 2026 historical quote", "publisher": "Nasdaq", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "market-avgo-jul24", "title": "Broadcom July 24, 2026 historical quote", "publisher": "Nasdaq", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "nvidia-sk-ai-factory-jul24", "title": "SK Group and NVIDIA Expand Strategic Partnership Across AI Factories and Next-Generation Memory", "publisher": "NVIDIA Newsroom", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "nvidia-korea-summit-jul24", "title": "At AI Summit, South Korea Outlines Its AI Future With NVIDIA and Partners", "publisher": "NVIDIA Blog", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "nvidia-kaist-lab-jul23", "title": "NVIDIA and KAIST Launch Joint AI Research Lab to Accelerate AI Innovation in Korea", "publisher": "NVIDIA Newsroom", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "wsj-openai-ohio-backstop-jul26", "title": "Nvidia in Talks With OpenAI to Guarantee $250 Billion Financing for Data Center", "publisher": "The Wall Street Journal", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "reuters-openai-ohio-backstop-jul26", "title": "Nvidia in talks with OpenAI to guarantee $250 billion financing for data center, WSJ reports", "publisher": "Reuters", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "theinformation-openai-ohio-campus-jun09", "title": "OpenAI in Talks to Lease 10 Gigawatt Ohio Data Center with Backing From Nvidia", "publisher": "The Information", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "nvidia-open-secure-ai-alliance-jul27", "title": "Industry Leaders Unite in Open Secure AI Alliance for AI Safety and Security", "publisher": "NVIDIA Blog", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "nvidia-ssi-partnership-jul27", "title": "Ilya Sutskever’s Safe Superintelligence Inc. and NVIDIA Announce Long-Term Strategic Partnership", "publisher": "NVIDIA Newsroom", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "openai-work-frontier-jul27", "title": "How AI is expanding what people do at work", "publisher": "OpenAI", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "reuters-altman-huang-warner-jul27", "title": "OpenAI CEO Sam Altman and NVIDIA CEO Jensen Huang to meet Senator Mark Warner", "publisher": "Reuters", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "ap-us-market-close-jul27", "title": "How major US stock indexes fared Monday 7/27/2026", "publisher": "The Associated Press", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "market-nvda-jul27", "title": "NVIDIA July 27, 2026 historical quote", "publisher": "Investing.com", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "meta-blackrock-el-paso-jul28", "title": "Meta Announces New Strategic Venture with BlackRock to Develop Data Center in El Paso", "publisher": "Meta Investor Relations", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "pacing-frontier-jul28", "title": "Pacing the Frontier", "publisher": "Pacing the Frontier", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "mcp-releases-status-jul28", "title": "Releases · modelcontextprotocol/modelcontextprotocol", "publisher": "Model Context Protocol on GitHub", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "mcp-versioning-status-jul28", "title": "Versioning - Model Context Protocol", "publisher": "Model Context Protocol", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "fcc-robots-inverters-jul28", "title": "FCC bans humanoid robots from China, citing unacceptable risks", "publisher": "The Guardian", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "reuters-chip-rout-jul28", "title": "Chip rout", "publisher": "Reuters", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "ap-us-market-close-jul28", "title": "How major US stock indexes fared Tuesday 7/28/2026", "publisher": "The Associated Press", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "market-qcom-jul28", "title": "Qualcomm Inc. stock underperforms Tuesday when compared to competitors", "publisher": "MarketWatch", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "reuters-wall-street-fed-ai-jul29", "title": "Wall Street closes down sharply after Fed holds rates unchanged", "publisher": "Reuters", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "ap-us-market-close-jul29", "title": "How major US stock indexes fared Wednesday 7/29/2026", "publisher": "The Associated Press", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "meta-q2-event-jul29", "title": "Q2 2026 Earnings Call", "publisher": "Meta Investor Relations", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "microsoft-fy26-q4-event-jul29", "title": "Microsoft announces quarterly earnings release date", "publisher": "Microsoft Source", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "mcp-releases-jul29", "title": "Releases · modelcontextprotocol/modelcontextprotocol", "publisher": "Model Context Protocol on GitHub", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "mcp-versioning-jul29", "title": "Versioning - Model Context Protocol", "publisher": "Model Context Protocol", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "eu-ai-transparency-guidelines-jul20", "title": "Commission publishes guidelines on transparency obligations for providers and deployers of certain AI systems", "publisher": "European Commission", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "microsoft-project-perception-jul27", "title": "Rethinking security for the age of AI", "publisher": "Microsoft", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "market-spx-jul29", "title": "S&P 500 Index historical data", "publisher": "MarketWatch", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "market-dji-jul29", "title": "Dow Jones Industrial Average historical data", "publisher": "Yahoo Finance", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "market-ixic-jul29", "title": "Nasdaq Composite historical data", "publisher": "Yahoo Finance", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "reuters-amazon-aws-jul30", "title": "Amazon lifts investment plans after strong cloud sales; shares jump", "publisher": "Reuters", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "reuters-wall-street-microsoft-jul30", "title": "Wall Street ends sharply higher, lifted by soaring Microsoft", "publisher": "Reuters", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "ap-apple-q3-jul30", "title": "Strong iPhone sales propel Apple fiscal Q3 results past Wall Street expectations", "publisher": "The Associated Press", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "amazon-q2-event-jul30", "title": "Q2 2026 Amazon.com earnings conference call", "publisher": "Amazon Investor Relations", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "mcp-releases-jul30", "title": "Releases · modelcontextprotocol/modelcontextprotocol", "publisher": "Model Context Protocol on GitHub", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "mcp-versioning-jul30", "title": "Versioning - Model Context Protocol 2026-07-28", "publisher": "Model Context Protocol", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "market-spx-jul30", "title": "S&P 500 Index historical data", "publisher": "MarketWatch", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "market-dji-jul30", "title": "Dow Jones Industrial Average historical data", "publisher": "Yahoo Finance", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "market-ixic-jul30", "title": "Nasdaq Composite historical data", "publisher": "Yahoo Finance", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "eu-article50-aug2", "title": "Commission publishes guidelines on transparency obligations for providers and deployers of certain AI systems", "publisher": "European Commission", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "eu-gpai-enforcement-aug2", "title": "Commission enforcement powers for obligations covering the most advanced general-purpose AI models", "publisher": "European Commission AI Act Service Desk", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "eu-ai-act-timeline-aug2", "title": "Timeline for the implementation of the EU AI Act", "publisher": "European Commission AI Act Service Desk", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "eu-governance-enforcement-aug2", "title": "Governance and enforcement of the AI Act", "publisher": "European Commission", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "huggingface-agent-incident-jul16", "title": "Security incident disclosure — July 2026", "publisher": "Hugging Face", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "reuters-eu-agent-talks-jul31", "title": "EU in talks with OpenAI, Anthropic after rogue AI agent hacks", "publisher": "Reuters", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "reuters-us-voluntary-tests-aug3", "title": "Meta, Anthropic, Google, OpenAI to meet Trump officials about AI safety testing", "publisher": "Reuters", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "whitehouse-frontier-security-order-jun2", "title": "Promoting Advanced Artificial Intelligence Innovation and Security", "publisher": "The White House", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "reuters-uk-voluntary-regulation-aug3", "title": "Britain says it is open to AI regulation if voluntary safeguards fall short", "publisher": "Reuters", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "uk-ai-cyber-code-jul10", "title": "Code of Practice for the Cyber Security of AI", "publisher": "UK Government", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "reuters-palantir-q2-aug3", "title": "Palantir lifts annual revenue forecast on steady demand for AI-powered data analytics", "publisher": "Reuters", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "ap-us-markets-aug3", "title": "How major US stock indexes fared Monday 8/3/2026", "publisher": "The Associated Press", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "reuters-ai-futures-aug4", "title": "US stock futures advance as strong AI-linked earnings continue", "publisher": "Reuters", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "market-spx-aug3", "title": "S&P 500 Index historical data", "publisher": "MarketWatch", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "market-dji-aug3", "title": "Dow Jones Industrial Average historical data", "publisher": "Yahoo Finance", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "market-ixic-aug3", "title": "Nasdaq Composite historical data", "publisher": "Yahoo Finance", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "reuters-whitehouse-open-weight-aug4", "title": "Trump advisers tell AI firms they will not safety-test open-weight models", "publisher": "Reuters", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "aisi-unsanctioned-agent-report-aug4", "title": "Incident Report: unsanctioned agent behaviour during cyber testing", "publisher": "UK AI Security Institute", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "openai-third-party-cyber-evals-aug4", "title": "Third-party cyber evaluations involving OpenAI models", "publisher": "OpenAI", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "reuters-agent-breaches-aug5", "title": "OpenAI, Anthropic AI agents implicated in new security breaches", "publisher": "Reuters", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "reuters-anthropic-global-affairs-aug4", "title": "Anthropic names global affairs chief to tackle AI policy as Trump tensions persist", "publisher": "Reuters", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "reuters-us-markets-aug4", "title": "Dow, S&P 500 close at record on AI-linked earnings, Mideast deal hopes", "publisher": "Reuters", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "market-spx-aug4", "title": "S&P 500 Index historical data — August 4, 2026 close", "publisher": "MarketWatch", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "market-dji-aug4", "title": "Dow Jones Industrial Average historical data — August 4, 2026 close", "publisher": "Yahoo Finance", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "market-ixic-aug4", "title": "Nasdaq Composite historical data — August 4, 2026 close", "publisher": "Yahoo Finance", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "reuters-google-ai-leadership-aug5", "title": "Google shakes up AI leadership as DeepMind chief shifts role", "publisher": "Reuters", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "reuters-meta-cyber-eval-aug5", "title": "Meta AI model hacks another company during testing", "publisher": "Reuters", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "reuters-amd-ai-payoff-aug5", "title": "AMD falls as investors demand bigger AI payoff", "publisher": "Reuters", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "reuters-global-markets-aug5", "title": "Stocks mixed with Nasdaq down; US oil eases on Iran talks", "publisher": "Reuters", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "market-spx-aug5", "title": "S&P 500 Index historical data — August 5, 2026 close", "publisher": "MarketWatch", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "market-dji-aug5", "title": "Dow Jones Industrial Average historical data — August 5, 2026 close", "publisher": "Yahoo Finance", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "market-ixic-aug5", "title": "Nasdaq Composite historical data — August 5, 2026 close", "publisher": "Yahoo Finance", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "reuters-alibaba-qwen38-license-aug7", "title": "Alibaba plans to charge big users of its next open-source AI model, sources say", "publisher": "Reuters", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "reuters-amd-taalas-aug6", "title": "AMD deepens AI inference bet with Taalas deal as chip race heats up", "publisher": "Reuters", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "reuters-deepseek-unitree-aug6", "title": "DeepSeek invests $20.8 million in Unitree's Shanghai IPO", "publisher": "Reuters", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "reuters-deepseek-funding-aug6", "title": "DeepSeek resumes funding round seeking nearly $8 billion, Bloomberg News reports", "publisher": "Reuters", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "reuters-us-ai-policy-pressure-aug6", "title": "Trump's tech ties come under bipartisan fire after AI agents go rogue", "publisher": "Reuters", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "reuters-us-markets-aug6", "title": "Stocks close lower as investors eye Mideast talks, earnings", "publisher": "Reuters", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "reuters-global-markets-aug7", "title": "Asian shares pause for US jobs, oil extends gains on Mideast risk", "publisher": "Reuters", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "market-spx-aug6", "title": "S&P 500 Index historical data — August 6, 2026 close", "publisher": "MarketWatch", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "market-dji-aug6", "title": "Dow Jones Industrial Average historical data — August 6, 2026 close", "publisher": "Yahoo Finance", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "market-ixic-aug6", "title": "Nasdaq Composite historical data — August 6, 2026 close", "publisher": "Yahoo Finance", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "whitehouse-frontier-ai-eo-jun2", "title": "Promoting Advanced Artificial Intelligence Innovation and Security", "publisher": "The White House", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "guardian-whitehouse-ai-framework-aug7", "title": "The White House's plan to vet potentially dangerous AI is cloaked in secrecy", "publisher": "The Guardian", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "reuters-open-weight-testing-aug4", "title": "Trump advisers tell AI firms they will not safety-test open-weight models", "publisher": "Reuters", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "reuters-us-jobs-aug7", "title": "US suffers unexpected job losses in July, markets dial back rate hike expectations", "publisher": "Reuters", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "reuters-us-markets-aug7", "title": "S&P closes at record high as soft jobs report eases rate-hike concerns", "publisher": "Reuters", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "reuters-inflation-weekahead-aug7", "title": "Wall St Week Ahead: Inflation data to test record-setting US stocks, Fed rate views", "publisher": "Reuters", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "wsj-source-foundry-aug7", "title": "Situational Awareness Bets $400 Million on Stealth Chip Startup After Crash", "publisher": "The Wall Street Journal", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "market-dji-aug7", "title": "Dow Jones Industrial Average historical data — August 7, 2026 close", "publisher": "Yahoo Finance", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "market-ixic-aug7", "title": "Nasdaq Composite historical data — August 7, 2026 close", "publisher": "Yahoo Finance", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "market-spx-aug7", "title": "S&P 500 Index historical data — August 7, 2026 close", "publisher": "MarketWatch", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "reuters-apple-qwen-mac-aug8", "title": "Apple says Mac users in China can connect to Alibaba's Qwen AI service", "publisher": "Reuters", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "reuters-openai-astra-cyber-aug7", "title": "OpenAI flags possible critical cybersecurity risk in upcoming model, tightens controls", "publisher": "Reuters", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "openai-preparedness-framework-v2", "title": "Preparedness Framework v2", "publisher": "OpenAI", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "reuters-aisi-agent-evals-aug5", "title": "OpenAI, Anthropic AI agents implicated in new security breaches", "publisher": "Reuters", "claimCount": 10, "criticalClaimCount": 10, "numericClaimCount": 9, "claimIds": [ "claim:benchmark.disclosure", "claim:benchmark.items.0", "claim:benchmark.items.1", "claim:benchmark.items.2", "claim:benchmark.items.3", "claim:benchmark.items.4", "claim:benchmark.items.5", "claim:benchmark.items.6", "claim:benchmark.items.7", "claim:dailyDelta.unchanged.0" ], "orphaned": false }, { "sourceId": "reuters-nist-ai-evals-aug7", "title": "Trump says Congress wants to regulate AI industry 'out of business'", "publisher": "Reuters", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "apple-cn-intelligence-support-aug8", "title": "Use Apple Intelligence on Mac", "publisher": "Apple Support China", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "openai-atlas-retirement-aug9", "title": "Evolving Atlas into ChatGPT for browser-based agentic work", "publisher": "OpenAI Help Center", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "anthropic-opus5-release-jul24", "title": "Claude Platform release notes — Claude Opus 5", "publisher": "Anthropic", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "arcprize-opus5-jul24", "title": "Claude Opus 5 — ARC-AGI results", "publisher": "ARC Prize", "claimCount": 10, "criticalClaimCount": 10, "numericClaimCount": 9, "claimIds": [ "claim:benchmark.disclosure", "claim:benchmark.items.0", "claim:benchmark.items.1", "claim:benchmark.items.2", "claim:benchmark.items.3", "claim:benchmark.items.4", "claim:benchmark.items.5", "claim:benchmark.items.6", "claim:benchmark.items.7", "claim:dailyDelta.unchanged.0" ], "orphaned": false }, { "sourceId": "arcprize-gpt56-jul9", "title": "GPT-5.6 — ARC-AGI results", "publisher": "ARC Prize", "claimCount": 10, "criticalClaimCount": 10, "numericClaimCount": 9, "claimIds": [ "claim:benchmark.disclosure", "claim:benchmark.items.0", "claim:benchmark.items.1", "claim:benchmark.items.2", "claim:benchmark.items.3", "claim:benchmark.items.4", "claim:benchmark.items.5", "claim:benchmark.items.6", "claim:benchmark.items.7", "claim:dailyDelta.unchanged.0" ], "orphaned": false }, { "sourceId": "reuters-qwen38-aug3", "title": "Alibaba unveils Qwen3.8-Max", "publisher": "Reuters", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "alibaba-model-catalog-aug9", "title": "Alibaba Cloud Model Studio model catalog", "publisher": "Alibaba Cloud", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "alibaba-model-pricing-aug9", "title": "Alibaba Cloud Model Studio pricing", "publisher": "Alibaba Cloud", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "deepseek-v4-apr24", "title": "DeepSeek V4-Pro and V4-Flash release notes", "publisher": "DeepSeek", "claimCount": 1, "criticalClaimCount": 1, "numericClaimCount": 1, "claimIds": [ "claim:models.26" ], "orphaned": false }, { "sourceId": "deepseek-v4-pricing-aug9", "title": "DeepSeek API pricing — V4-Flash", "publisher": "DeepSeek", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "artificialanalysis-v4flash-aug9", "title": "DeepSeek V4 Flash — independent model analysis", "publisher": "Artificial Analysis", "claimCount": 10, "criticalClaimCount": 10, "numericClaimCount": 9, "claimIds": [ "claim:benchmark.disclosure", "claim:benchmark.items.0", "claim:benchmark.items.1", "claim:benchmark.items.2", "claim:benchmark.items.3", "claim:benchmark.items.4", "claim:benchmark.items.5", "claim:benchmark.items.6", "claim:benchmark.items.7", "claim:dailyDelta.unchanged.0" ], "orphaned": false }, { "sourceId": "reuters-deepseek-v4-aug3", "title": "DeepSeek V4-Flash cost and performance reporting", "publisher": "Reuters", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "reuters-bigtech-leases-aug4", "title": "AI data-centre race builds $1 trillion lease burden for Big Tech", "publisher": "Reuters", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "sec-msft-leases-mar31", "title": "Microsoft Form 10-Q — leases not yet commenced", "publisher": "U.S. Securities and Exchange Commission", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "sec-meta-leases-mar31", "title": "Meta Platforms Form 10-Q lease commitments", "publisher": "U.S. Securities and Exchange Commission", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "sec-alphabet-leases-mar31", "title": "Alphabet Form 10-Q — leases not yet commenced", "publisher": "U.S. Securities and Exchange Commission", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "aisi-test-time-compute", "title": "Measuring AI agents progress on multi-step cyber attack scenarios", "publisher": "UK AI Security Institute", "claimCount": 10, "criticalClaimCount": 10, "numericClaimCount": 9, "claimIds": [ "claim:benchmark.disclosure", "claim:benchmark.items.0", "claim:benchmark.items.1", "claim:benchmark.items.2", "claim:benchmark.items.3", "claim:benchmark.items.4", "claim:benchmark.items.5", "claim:benchmark.items.6", "claim:benchmark.items.7", "claim:dailyDelta.unchanged.0" ], "orphaned": false }, { "sourceId": "anthropic-models-overview-aug9", "title": "Claude models overview", "publisher": "Anthropic", "claimCount": 1, "criticalClaimCount": 1, "numericClaimCount": 1, "claimIds": [ "claim:models.19" ], "orphaned": false }, { "sourceId": "reuters-nvidia-financing-aug10", "title": "Wall Street giants partner with Nvidia on $500 billion AI financing plan", "publisher": "Reuters", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "nvidia-naver-brookfield-jul24", "title": "NAVER, NVIDIA and Brookfield to Expand Korea’s National AI Factory Infrastructure Buildout", "publisher": "NVIDIA Newsroom", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "reuters-datacenter-lender-risk-aug10", "title": "Lenders scrutinize U.S. data-center financing as community opposition builds", "publisher": "Reuters", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "reuters-house-agent-oversight-aug10", "title": "U.S. House Democrats press Anthropic, OpenAI about AI-agent incidents", "publisher": "Reuters", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "openai-hf-incident-jul21", "title": "Hugging Face model-evaluation security incident", "publisher": "OpenAI", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "huggingface-agent-intrusion-jul27", "title": "Agent intrusion technical timeline", "publisher": "Hugging Face", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "anthropic-containment-may25", "title": "How we contain Claude across products", "publisher": "Anthropic", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "reuters-meta-glimmer-aug10", "title": "Meta launches new AI model as Zuckerberg champions open-weight push", "publisher": "Reuters", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "reuters-maia300-aug10", "title": "Microsoft plans to unveil its new Maia 300 AI chip this fall, The Information reports", "publisher": "Reuters", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "microsoft-maia300-transcript-mar4", "title": "Morgan Stanley Technology, Media & Telecom Conference", "publisher": "Microsoft Investor Relations", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "cvpr-agentsafe-2026", "title": "AGENTSAFE: Benchmarking the Safety of Embodied Agents on Hazardous Instructions", "publisher": "CVPR 2026", "claimCount": 10, "criticalClaimCount": 10, "numericClaimCount": 9, "claimIds": [ "claim:benchmark.disclosure", "claim:benchmark.items.0", "claim:benchmark.items.1", "claim:benchmark.items.2", "claim:benchmark.items.3", "claim:benchmark.items.4", "claim:benchmark.items.5", "claim:benchmark.items.6", "claim:benchmark.items.7", "claim:dailyDelta.unchanged.0" ], "orphaned": false }, { "sourceId": "acl-safeagent-2026", "title": "SafeAgent: Safeguarding LLM Agents via an Automated Risk Simulator", "publisher": "ACL 2026", "claimCount": 10, "criticalClaimCount": 10, "numericClaimCount": 9, "claimIds": [ "claim:benchmark.disclosure", "claim:benchmark.items.0", "claim:benchmark.items.1", "claim:benchmark.items.2", "claim:benchmark.items.3", "claim:benchmark.items.4", "claim:benchmark.items.5", "claim:benchmark.items.6", "claim:benchmark.items.7", "claim:dailyDelta.unchanged.0" ], "orphaned": false }, { "sourceId": "ap-us-markets-aug10", "title": "How major U.S. stock indexes fared Monday 8/10/2026", "publisher": "Associated Press", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "reuters-coreweave-supermicro-aug12", "title": "CoreWeave, Super Micro surge on signs of sustained AI buildout", "publisher": "Reuters", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "coreweave-q1-backlog-definition-2026", "title": "CoreWeave Reports Strong First Quarter 2026 Results", "publisher": "CoreWeave Investor Relations", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "supermicro-q4-preliminary-jul21", "title": "Supermicro Provides Fourth Quarter of Fiscal Year 2026 Preliminary Business Update", "publisher": "Super Micro Computer Investor Relations", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "reuters-cisco-ai-orders-aug12", "title": "Cisco forecasts annual revenue above estimates on sustained AI spending", "publisher": "Reuters", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "reuters-foxconn-cowos-aug12", "title": "Taiwan’s Foxconn reports 35% rise in Q2 profit on AI demand, beats forecasts", "publisher": "Reuters", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "tsmc-cowos-3dfabric", "title": "3DFabric: CoWoS and advanced packaging", "publisher": "TSMC", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "reuters-taiwan-ai-attack-aug13", "title": "Taiwan says it was targeted last month in AI-driven hacking campaign", "publisher": "Reuters", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "ft-taiwan-agent-cyber-aug12", "title": "China-linked hackers hit Taiwan in unprecedented autonomous AI cyber attack", "publisher": "Financial Times", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "taiwan-openclaw-advisory-mar2026", "title": "OpenClaw cybersecurity risk advisory", "publisher": "Taiwan Administration for Cyber Security", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "reuters-google-deepmind-aug12", "title": "Inside the Google executive moves that led to its big AI reshuffle", "publisher": "Reuters", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "google-gemini-35-current", "title": "Gemini 3.5", "publisher": "Google DeepMind", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "reuters-cerebras-q2-aug12", "title": "Cerebras raises annual targets on strong AI chip demand", "publisher": "Reuters", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "reuters-lightcap-openai-aug11", "title": "Senior OpenAI executive Brad Lightcap to leave for new venture", "publisher": "Reuters", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "reuters-japan-ai-adoption-aug12", "title": "Strong majority of Japanese firms have yet to fully embrace AI", "publisher": "Reuters / Nikkei Research", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "whitehouse-frontier-order-jun2026", "title": "Promoting Advanced Artificial Intelligence Innovation and Security", "publisher": "The White House", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "whitehouse-nspm11-jun2026", "title": "National Security Presidential Memorandum 11", "publisher": "The White House", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "wired-open-model-policy-aug12", "title": "The White House Is Going to Expand Its AI Policy", "publisher": "WIRED", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "ap-us-markets-aug12", "title": "How major U.S. stock indexes fared Wednesday 8/12/2026", "publisher": "Associated Press", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "reuters-deepseek-v4-0813-aug13", "title": "DeepSeek releases official V4 Pro model as it steps up expansion", "publisher": "Reuters", "claimCount": 1, "criticalClaimCount": 1, "numericClaimCount": 1, "claimIds": [ "claim:models.18" ], "orphaned": false }, { "sourceId": "reuters-deepseek-pricing-aug13", "title": "DeepSeek raises API pricing for its V4 models", "publisher": "Reuters", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "deepseek-pricing-current-aug13", "title": "Models & Pricing", "publisher": "DeepSeek API Docs", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "deepseek-v4-preview-apr24", "title": "DeepSeek-V4-Pro and DeepSeek-V4-Flash release notes", "publisher": "DeepSeek API Docs", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "reuters-databricks-aug13", "title": "Databricks raises $5 billion in financing at $190 billion valuation", "publisher": "Reuters", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "reuters-databricks-jul17", "title": "Databricks valued at $188 billion after Coatue-led investment term sheet", "publisher": "Reuters", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "reuters-applied-materials-aug13", "title": "Applied Materials forecasts quarterly revenue above estimates", "publisher": "Reuters", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "reuters-smic-aug13", "title": "SMIC profit more than triples on AI-driven chip demand", "publisher": "Reuters", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "reuters-lt-together-aug13", "title": "India’s Larsen & Toubro secures AI data-centre order worth up to $1.57 billion", "publisher": "Reuters", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "axios-openai-cro-aug13", "title": "OpenAI revenue chief Denise Dresser to leave", "publisher": "Axios", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "openai-enterprise-apr8", "title": "The next phase of enterprise AI", "publisher": "OpenAI", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "reuters-amd-debt-aug13", "title": "AMD looks to raise $4 billion to $5 billion in debt offering", "publisher": "Reuters", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "arxiv-runtime-contract-aug11", "title": "Agent Safety Should Be a Runtime Contract", "publisher": "arXiv", "claimCount": 10, "criticalClaimCount": 10, "numericClaimCount": 9, "claimIds": [ "claim:benchmark.disclosure", "claim:benchmark.items.0", "claim:benchmark.items.1", "claim:benchmark.items.2", "claim:benchmark.items.3", "claim:benchmark.items.4", "claim:benchmark.items.5", "claim:benchmark.items.6", "claim:benchmark.items.7", "claim:dailyDelta.unchanged.0" ], "orphaned": false }, { "sourceId": "ap-us-markets-aug13", "title": "How major U.S. stock indexes fared Thursday 8/13/2026", "publisher": "Associated Press", "claimCount": 3, "criticalClaimCount": 3, "numericClaimCount": 3, "claimIds": [ "claim:markets.0", "claim:markets.1", "claim:markets.2" ], "orphaned": false }, { "sourceId": "reuters-smic-pricing-aug14", "title": "Chinese chipmaker SMIC increases prices on strong AI demand", "publisher": "Reuters", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "reuters-private-credit-roundup-aug14", "title": "Private credit roundup: Nvidia's half trillion for chips financing, plus others", "publisher": "Reuters", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "reuters-goldman-nvidia-finance-aug14", "title": "Goldman in talks with investors on Nvidia financing deal after landing prized role, sources say", "publisher": "Reuters", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "reuters-zai-glm53-aug14", "title": "China's Z.ai says new model nears Anthropic's Mythos 5 in cyber-defence tests", "publisher": "Reuters", "claimCount": 10, "criticalClaimCount": 10, "numericClaimCount": 9, "claimIds": [ "claim:benchmark.disclosure", "claim:benchmark.items.0", "claim:benchmark.items.1", "claim:benchmark.items.2", "claim:benchmark.items.3", "claim:benchmark.items.4", "claim:benchmark.items.5", "claim:benchmark.items.6", "claim:benchmark.items.7", "claim:dailyDelta.unchanged.0" ], "orphaned": false }, { "sourceId": "anthropic-risk-report-aug14", "title": "Risk Report: August 2026", "publisher": "Anthropic", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "axios-anthropic-risk-aug14", "title": "Anthropic sees AI risks rising, no plan to release stronger Model 2", "publisher": "Axios", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "reuters-pax-silica-exclusivity-aug15", "title": "US to tell partners they must pick sides in AI race with China", "publisher": "Reuters", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "vietnam-decision33-moj-aug15", "title": "Ban hành Danh mục hệ thống trí tuệ nhân tạo có rủi ro cao", "publisher": "Vietnam Ministry of Justice", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "vietnam-decision33-official-aug15", "title": "Quyết định số 33/2026/QĐ-TTg: Ban hành Danh mục hệ thống trí tuệ nhân tạo có rủi ro cao", "publisher": "Government of Vietnam", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "microsoft-foundry-retirement-aug15", "title": "Model retirement schedule - Microsoft Foundry", "publisher": "Microsoft Learn", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "sap-analytics-cloud-aug15", "title": "What's New in SAP Analytics Cloud", "publisher": "SAP Help Portal", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "arxiv-multi-agent-institutions-aug10", "title": "Multi-Agent AI Safety as an Institutional Design Problem", "publisher": "arXiv", "claimCount": 10, "criticalClaimCount": 10, "numericClaimCount": 9, "claimIds": [ "claim:benchmark.disclosure", "claim:benchmark.items.0", "claim:benchmark.items.1", "claim:benchmark.items.2", "claim:benchmark.items.3", "claim:benchmark.items.4", "claim:benchmark.items.5", "claim:benchmark.items.6", "claim:benchmark.items.7", "claim:dailyDelta.unchanged.0" ], "orphaned": false }, { "sourceId": "reuters-amat-expectations-aug14", "title": "Applied Materials slips as investors seek faster growth after stellar run", "publisher": "Reuters", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "deepseek-pricing-current-aug15", "title": "Models & Pricing", "publisher": "DeepSeek API Docs", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "reuters-anthropic-ipo-forecast-aug15", "title": "Anthropic IPO valuation hinges on $190-200 billion 2028 revenue forecast, sources say", "publisher": "Reuters", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "reuters-nvidia-openai-ohio-backstop-aug14", "title": "Nvidia scales back funding guarantee for Ohio OpenAI data center, WSJ reports", "publisher": "Reuters", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "reuters-uscc-data-aug18", "title": "US advisory body says China’s data dominance gives it AI advantage", "publisher": "Reuters", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "uscc-two-loops-mar23", "title": "Two Loops: How China’s Open AI Strategy Reinforces Its Industrial Dominance", "publisher": "U.S.-China Economic and Security Review Commission", "claimCount": 10, "criticalClaimCount": 10, "numericClaimCount": 9, "claimIds": [ "claim:benchmark.disclosure", "claim:benchmark.items.0", "claim:benchmark.items.1", "claim:benchmark.items.2", "claim:benchmark.items.3", "claim:benchmark.items.4", "claim:benchmark.items.5", "claim:benchmark.items.6", "claim:benchmark.items.7", "claim:dailyDelta.unchanged.0" ], "orphaned": false }, { "sourceId": "china-nda-data-survey-apr29", "title": "全国数据资源调查报告（2025年）正式发布", "publisher": "China National Data Administration", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "reuters-china-robots-aug18", "title": "Beyond marathons and backflips, China’s robots face a commercial test", "publisher": "Reuters", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "reuters-unitree-us-research-aug18", "title": "How US military funding propelled China’s robot dogs", "publisher": "Reuters", "claimCount": 10, "criticalClaimCount": 10, "numericClaimCount": 9, "claimIds": [ "claim:benchmark.disclosure", "claim:benchmark.items.0", "claim:benchmark.items.1", "claim:benchmark.items.2", "claim:benchmark.items.3", "claim:benchmark.items.4", "claim:benchmark.items.5", "claim:benchmark.items.6", "claim:benchmark.items.7", "claim:dailyDelta.unchanged.0" ], "orphaned": false }, { "sourceId": "ap-chatgpt-teens-aug18", "title": "OpenAI launches ChatGPT for Teens, promising a more age-appropriate chatbot", "publisher": "Associated Press", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "openai-age-prediction-aug18", "title": "Age prediction in ChatGPT", "publisher": "OpenAI Help Center", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "openai-parental-controls-aug18", "title": "Parental Controls on ChatGPT - FAQ", "publisher": "OpenAI Help Center", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "reuters-nvidia-ohio-aug17", "title": "Nvidia to provide up to $105 billion guarantee for OpenAI’s Ohio data center", "publisher": "Reuters", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "reuters-baidu-q2-aug18", "title": "Baidu revenue misses as advertising slide outweighs AI cloud gains", "publisher": "Reuters", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "baidu-ir-q2-call-aug18", "title": "Q2 2026 Baidu Earnings Conference Call", "publisher": "Baidu Investor Relations", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "reuters-anthropic-run-rate-aug17", "title": "Anthropic revenue run rate tops $65 billion, source says", "publisher": "Reuters", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "deepseek-pricing-current-aug18", "title": "Models & Pricing", "publisher": "DeepSeek API Docs", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "reuters-openai-slowdown-aug18", "title": "OpenAI slows model training to bolster security after Hugging Face hack", "publisher": "Reuters", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "openai-astra-critical-aug7", "title": "Responding to the next frontier of critical cyber capabilities", "publisher": "OpenAI", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "openai-third-party-cyber-aug4", "title": "Third-party cyber evaluations involving OpenAI models", "publisher": "OpenAI", "claimCount": 10, "criticalClaimCount": 10, "numericClaimCount": 9, "claimIds": [ "claim:benchmark.disclosure", "claim:benchmark.items.0", "claim:benchmark.items.1", "claim:benchmark.items.2", "claim:benchmark.items.3", "claim:benchmark.items.4", "claim:benchmark.items.5", "claim:benchmark.items.6", "claim:benchmark.items.7", "claim:dailyDelta.unchanged.0" ], "orphaned": false }, { "sourceId": "openai-preparedness-apr15", "title": "Our updated Preparedness Framework", "publisher": "OpenAI", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "reuters-cerebras-cs4-aug19", "title": "Cerebras launches new server chip and system designed to speed AI chatbots", "publisher": "Reuters", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "jll-emea-dc-aug3", "title": "EMEA data centre mid-year 2026 report", "publisher": "JLL", "claimCount": 10, "criticalClaimCount": 10, "numericClaimCount": 9, "claimIds": [ "claim:benchmark.disclosure", "claim:benchmark.items.0", "claim:benchmark.items.1", "claim:benchmark.items.2", "claim:benchmark.items.3", "claim:benchmark.items.4", "claim:benchmark.items.5", "claim:benchmark.items.6", "claim:benchmark.items.7", "claim:dailyDelta.unchanged.0" ], "orphaned": false }, { "sourceId": "reuters-europe-dc-aug19", "title": "Europe AI data centres seek cheaper, quicker energy and land", "publisher": "Reuters", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "reuters-china-sovereignty-aug19", "title": "China urges respect for digital sovereignty in AI race", "publisher": "Reuters", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "reuters-us-pick-sides-aug14", "title": "US to tell partners they must pick sides in AI race with China", "publisher": "Reuters", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "china-mfa-digital-sovereignty-jul23", "title": "Foreign Ministry Spokesperson’s Remarks on China Releasing the Position Paper on Global Cyber Governance in the Digital Intelligent Age", "publisher": "Ministry of Foreign Affairs of the People’s Republic of China", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "reuters-guidelight-control-aug19", "title": "AI firms can’t yet contain what they’ve built, study finds", "publisher": "Reuters", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "guidelight-control-standard-v1", "title": "Control v1.0", "publisher": "Guidelight AI Standards", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "guidelight-development-process", "title": "Our Standards Development Process", "publisher": "Guidelight AI Standards", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "aisi-unsanctioned-agent-aug4", "title": "Incident Report: unsanctioned agent behaviour during cyber testing", "publisher": "UK AI Security Institute", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "reuters-aisi-maintainer-aug20", "title": "How a Texas student blew the whistle on a rogue AI hacking attempt", "publisher": "Reuters", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "reuters-marvell-google-aug19", "title": "Marvell gives Google option to buy $12.2 billion stake in custom AI chip deal", "publisher": "Reuters", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "axios-openai-private-safety-aug19", "title": "OpenAI previews zero-retention safety system as Anthropic requires data logs", "publisher": "Axios", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "reuters-brazil-sovereign-ai-aug20", "title": "Brazil launches AI supercomputer push, splits projects between Chinese, US firms", "publisher": "Reuters", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "reuters-nvidia-cloverleaf-aug21", "title": "Nvidia invests in data center developer Cloverleaf Infrastructure", "publisher": "Reuters", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "reuters-anthropic-retention-aug20", "title": "Anthropic plans to change enterprise data retention policy, source says", "publisher": "Reuters", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "anthropic-covered-model-retention", "title": "Data retention practices for Covered Models", "publisher": "Anthropic / Claude Help Center", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "reuters-ai-financing-yields-aug21", "title": "Wall St Week Ahead: Nvidia earnings, Jackson Hole to test pillars of stock rally", "publisher": "Reuters", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "nvidia-financing-platforms-aug10", "title": "NVIDIA Partners With Apollo, BlackRock, Blackstone, Brookfield, Goldman Sachs and KKR to Establish AI Compute Infrastructure Financing Platforms to Mobilize Over $500 Billion of Third-Party Capital", "publisher": "NVIDIA Investor Relations", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "reuters-openai-sol-price-aug21", "title": "OpenAI cuts developer pricing for frontier GPT-5.6 Sol model by more than 20%", "publisher": "Reuters", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "openai-api-pricing-aug22", "title": "API pricing", "publisher": "OpenAI", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "reuters-india-it-contracts-aug21", "title": "AI reshapes India's IT services sector contracts as clients demand more for less", "publisher": "Reuters", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "reuters-aisi-agent-aug20", "title": "How a Texas student blew the whistle on a rogue AI hacking attempt", "publisher": "Reuters", "claimCount": 10, "criticalClaimCount": 10, "numericClaimCount": 9, "claimIds": [ "claim:benchmark.disclosure", "claim:benchmark.items.0", "claim:benchmark.items.1", "claim:benchmark.items.2", "claim:benchmark.items.3", "claim:benchmark.items.4", "claim:benchmark.items.5", "claim:benchmark.items.6", "claim:benchmark.items.7", "claim:dailyDelta.unchanged.0" ], "orphaned": false }, { "sourceId": "aisi-agent-incident-report", "title": "Incident Report: unsanctioned agent behaviour during cyber testing", "publisher": "UK AI Security Institute", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "reuters-ace-robotics-aug21", "title": "ACE Robotics chairman says robot brains will have ChatGPT moment by end of 2027", "publisher": "Reuters", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "kairos-open-source-repo", "title": "Kairos: official code for a cross-embodiment world model", "publisher": "Kairos-AGI / GitHub", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "cltr-loss-of-control-report-2026", "title": "Loss of Control Observatory 2026 Telemetry Report", "publisher": "Centre for Long-Term Resilience / UK AISI", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "industry-cyber-defense-call-2026", "title": "A Call for Collective Action on Cyber Defense", "publisher": "OpenAI, Anthropic, Google, Microsoft, AWS, Nvidia Coalition", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "zhipu-glm53-flash-release", "title": "GLM-5.3-Flash open weights release and pricing", "publisher": "Z.ai / Hugging Face", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "alibaba-qwen38-flash-docs", "title": "Qwen3.8-Flash-Next API and architecture documentation", "publisher": "Alibaba Cloud", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "cisco-agent-rollout-2026", "title": "Cisco deploys MyAgent to 90,000 employees and expands Secure AI Factory", "publisher": "Cisco Systems", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "nist-eu-agent-sandboxing-2026", "title": "Joint Guidelines on Autonomous Agent Sandboxing and Ephemeral Credentials", "publisher": "NIST & European AI Office", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "cloud-asic-inference-economics", "title": "Q3 2026 Cloud Datacenter Silicon TCO and Inference Economics", "publisher": "AWS & Google Cloud Infrastructure", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "arc-anthropic-cot-faithfulness", "title": "Evaluating Chain-of-Thought Faithfulness Under Outcome Reinforcement Learning", "publisher": "Anthropic Research & Alignment Research Center", "claimCount": 10, "criticalClaimCount": 10, "numericClaimCount": 9, "claimIds": [ "claim:benchmark.disclosure", "claim:benchmark.items.0", "claim:benchmark.items.1", "claim:benchmark.items.2", "claim:benchmark.items.3", "claim:benchmark.items.4", "claim:benchmark.items.5", "claim:benchmark.items.6", "claim:benchmark.items.7", "claim:dailyDelta.unchanged.0" ], "orphaned": false }, { "sourceId": "ollama-streaming-audio-quant", "title": "4-bit Streaming KV Audio Quantization Kernels for Edge Devices", "publisher": "llama.cpp / Ollama Open Source", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "llama-3-4-moe-release", "title": "Meta Llama 3.4-MoE Weights Release & Speculative Draft Heads", "publisher": "Meta AI / Hugging Face", "claimCount": 10, "criticalClaimCount": 10, "numericClaimCount": 9, "claimIds": [ "claim:benchmark.disclosure", "claim:benchmark.items.0", "claim:benchmark.items.1", "claim:benchmark.items.2", "claim:benchmark.items.3", "claim:benchmark.items.4", "claim:benchmark.items.5", "claim:benchmark.items.6", "claim:benchmark.items.7", "claim:dailyDelta.unchanged.0" ], "orphaned": false }, { "sourceId": "ferc-doe-grid-colocation-order", "title": "FERC Docket RM26-4-000 & DOE Joint Order on 500MW+ Datacenter Co-Location", "publisher": "Federal Energy Regulatory Commission & US DOE", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "clean-energy-ppa-economics", "title": "Q3 2026 Hyperscaler Energy PPA Filings and Clean Baseload Economics", "publisher": "AWS & Google Cloud Infrastructure Telemetry", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "stanford-berkeley-steering", "title": "Deterministic Containment of Autonomous Agent Tool Invocation via Residual Activation Steering", "publisher": "Stanford CRFM & UC Berkeley AI Research", "claimCount": 10, "criticalClaimCount": 10, "numericClaimCount": 9, "claimIds": [ "claim:benchmark.disclosure", "claim:benchmark.items.0", "claim:benchmark.items.1", "claim:benchmark.items.2", "claim:benchmark.items.3", "claim:benchmark.items.4", "claim:benchmark.items.5", "claim:benchmark.items.6", "claim:benchmark.items.7", "claim:dailyDelta.unchanged.0" ], "orphaned": false }, { "sourceId": "bird-sql-evaluation-harness", "title": "BIRD-SQL Cross-Domain Enterprise Relational Database Benchmark Leaderboard", "publisher": "BIRD-SQL Consortium / DAMO", "claimCount": 10, "criticalClaimCount": 10, "numericClaimCount": 9, "claimIds": [ "claim:benchmark.disclosure", "claim:benchmark.items.0", "claim:benchmark.items.1", "claim:benchmark.items.2", "claim:benchmark.items.3", "claim:benchmark.items.4", "claim:benchmark.items.5", "claim:benchmark.items.6", "claim:benchmark.items.7", "claim:dailyDelta.unchanged.0" ], "orphaned": false }, { "sourceId": "google-gemini-38-launch", "title": "Gemini 3.8 Architecture, Multimodal Latency, and Reasoning Evaluation", "publisher": "Google DeepMind", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "openai-astra-safety-audit", "title": "OpenAI Astra Preparedness Evaluation and ChatGPT Enterprise Rollout", "publisher": "OpenAI", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "mistral-large-3-release", "title": "Mistral Large 3 Architecture Whitepaper and Open Weights", "publisher": "Mistral AI", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "kv-cache-mla-systems", "title": "Multi-Head Latent Attention and 4-Bit PagedAttention Serving Implementations", "publisher": "vLLM Project & TensorRT-LLM", "claimCount": 10, "criticalClaimCount": 10, "numericClaimCount": 9, "claimIds": [ "claim:benchmark.disclosure", "claim:benchmark.items.0", "claim:benchmark.items.1", "claim:benchmark.items.2", "claim:benchmark.items.3", "claim:benchmark.items.4", "claim:benchmark.items.5", "claim:benchmark.items.6", "claim:benchmark.items.7", "claim:dailyDelta.unchanged.0" ], "orphaned": false }, { "sourceId": "deepseek-kimi-long-context-pricing", "title": "DeepSeek and Moonshot Long-Context API Pricing Schedule", "publisher": "DeepSeek Open Platform & Moonshot AI", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "eu-ai-office-gpai-guidelines", "title": "European AI Office Implementation Rules for Systemic GPAI Models", "publisher": "European Commission AI Office", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "nist-ai-600-2-agent-profile", "title": "NIST AI 600-2: Profile for Assessing Autonomous Agent Deployments", "publisher": "US National Institute of Standards and Technology", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "mit-cmu-decomposed-semantic-inversion", "title": "Decomposed Semantic Inversion in Long-Context LLMs", "publisher": "MIT CSAIL & Carnegie Mellon University", "claimCount": 10, "criticalClaimCount": 10, "numericClaimCount": 9, "claimIds": [ "claim:benchmark.disclosure", "claim:benchmark.items.0", "claim:benchmark.items.1", "claim:benchmark.items.2", "claim:benchmark.items.3", "claim:benchmark.items.4", "claim:benchmark.items.5", "claim:benchmark.items.6", "claim:benchmark.items.7", "claim:dailyDelta.unchanged.0" ], "orphaned": false }, { "sourceId": "anthropic-fable51-release", "title": "Claude Fable 5.1 and Mythos 5.1 Architecture, Safeguards, and Production Pricing", "publisher": "Anthropic", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "swebench-pro-evaluation", "title": "SWE-bench Verified Saturation Analysis and SWE-bench Pro Evaluation Methodology", "publisher": "SWE-bench Consortium", "claimCount": 10, "criticalClaimCount": 10, "numericClaimCount": 9, "claimIds": [ "claim:benchmark.disclosure", "claim:benchmark.items.0", "claim:benchmark.items.1", "claim:benchmark.items.2", "claim:benchmark.items.3", "claim:benchmark.items.4", "claim:benchmark.items.5", "claim:benchmark.items.6", "claim:benchmark.items.7", "claim:dailyDelta.unchanged.0" ], "orphaned": false }, { "sourceId": "nvidia-rubin-nvl72-infrastructure", "title": "NVIDIA Rubin NVL72 and Blackwell Ultra Rack-Scale Architecture Specifications", "publisher": "NVIDIA Corporation", "claimCount": 10, "criticalClaimCount": 10, "numericClaimCount": 9, "claimIds": [ "claim:benchmark.disclosure", "claim:benchmark.items.0", "claim:benchmark.items.1", "claim:benchmark.items.2", "claim:benchmark.items.3", "claim:benchmark.items.4", "claim:benchmark.items.5", "claim:benchmark.items.6", "claim:benchmark.items.7", "claim:dailyDelta.unchanged.0" ], "orphaned": false }, { "sourceId": "anthropic-cache-pricing-api", "title": "Claude API Pricing Schedule and Context Caching Specifications", "publisher": "Anthropic", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "nist-ai-600-2-agent-sandboxing", "title": "NIST AI 600-2: Profile for Assessing Autonomous Agent Deployments", "publisher": "US National Institute of Standards and Technology", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "project-glasswing-safeguards", "title": "Project Glasswing Defensive Cybersecurity Verification Framework and Enterprise Frontier Safeguards", "publisher": "Anthropic & Coalition Partners", "claimCount": 10, "criticalClaimCount": 10, "numericClaimCount": 9, "claimIds": [ "claim:benchmark.disclosure", "claim:benchmark.items.0", "claim:benchmark.items.1", "claim:benchmark.items.2", "claim:benchmark.items.3", "claim:benchmark.items.4", "claim:benchmark.items.5", "claim:benchmark.items.6", "claim:benchmark.items.7", "claim:dailyDelta.unchanged.0" ], "orphaned": false }, { "sourceId": "openai-operator-enterprise-launch", "title": "OpenAI Operator Enterprise Preview Documentation and System Security Specifications", "publisher": "OpenAI", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "cmu-berkeley-agent-injection-audit", "title": "Indirect Prompt Injection and Tool Poisoning in Autonomous Multi-Turn Agents: Empirical Vulnerability Assessment and Structural Defenses", "publisher": "Carnegie Mellon University & UC Berkeley", "claimCount": 10, "criticalClaimCount": 10, "numericClaimCount": 9, "claimIds": [ "claim:benchmark.disclosure", "claim:benchmark.items.0", "claim:benchmark.items.1", "claim:benchmark.items.2", "claim:benchmark.items.3", "claim:benchmark.items.4", "claim:benchmark.items.5", "claim:benchmark.items.6", "claim:benchmark.items.7", "claim:dailyDelta.unchanged.0" ], "orphaned": false }, { "sourceId": "ferc-nuclear-baseload-filings", "title": "Federal Energy Regulatory Commission Docket on Data Center Co-Location and Baseload Interconnections", "publisher": "Federal Energy Regulatory Commission (FERC)", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "osworld-webarena-benchmark-audit", "title": "OSWorld and WebArena Autonomous Agent Evaluation Leaderboards and Dynamic Environment Test Logs", "publisher": "OSWorld / WebArena Evaluation Consortium", "claimCount": 10, "criticalClaimCount": 10, "numericClaimCount": 9, "claimIds": [ "claim:benchmark.disclosure", "claim:benchmark.items.0", "claim:benchmark.items.1", "claim:benchmark.items.2", "claim:benchmark.items.3", "claim:benchmark.items.4", "claim:benchmark.items.5", "claim:benchmark.items.6", "claim:benchmark.items.7", "claim:dailyDelta.unchanged.0" ], "orphaned": false }, { "sourceId": "eu-ai-office-gpai-code-draft", "title": "Draft General Purpose AI Code of Practice for Systemic Risk Frontier Models", "publisher": "European AI Office", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "agent-virtualization-unit-economics", "title": "Enterprise Autonomous Agent Infrastructure and Virtualization Cost Analysis", "publisher": "Cloud Infrastructure Research Group", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "deepseek-v35-release-report", "title": "DeepSeek-V3.5 Technical Report: Multi-Head Latent Attention v2, 671B Sparse MoE Architecture and Inference Economics", "publisher": "DeepSeek AI", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "us-bis-nist-1026-reporting-rule", "title": "Bureau of Industry and Security Rule: Mandatory Reporting Requirements for Frontier Dual-Use Foundation Models Exceeding 10^26 FLOPs (15 CFR Part 702)", "publisher": "US Department of Commerce & NIST", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "tsmc-fab20-n2-packaging-disclosure", "title": "TSMC Hsinchu Fab 20 2nm GAAFET Production Update and 2027 Advanced Packaging CoWoS-L Allocation Audit", "publisher": "TSMC / Taiwan Ministry of Economic Affairs", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "eth-berkeley-speculative-leak-paper", "title": "Speculative Leak: Microarchitectural Side-Channels and Cache Contention in Speculative Decoding Inference Engines (CVE-2026-38192)", "publisher": "ETH Zurich & UC Berkeley Systems and Security Lab", "claimCount": 10, "criticalClaimCount": 10, "numericClaimCount": 9, "claimIds": [ "claim:benchmark.disclosure", "claim:benchmark.items.0", "claim:benchmark.items.1", "claim:benchmark.items.2", "claim:benchmark.items.3", "claim:benchmark.items.4", "claim:benchmark.items.5", "claim:benchmark.items.6", "claim:benchmark.items.7", "claim:dailyDelta.unchanged.0" ], "orphaned": false }, { "sourceId": "enterprise-inference-tco-audit-sep26", "title": "Quarterly Enterprise AI Infrastructure Audit: The 40 Million Token Crossover from Public Cloud APIs to Dedicated On-Premises Clusters", "publisher": "Enterprise Cloud Infrastructure Research Group", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "anthropic-mcp2-claude-code-spec", "title": "Model Context Protocol 2.0 Specification & Cryptographic Attestation Architecture: Claude Code General Availability", "publisher": "Anthropic", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "virginia-deq-loudoun-water-directive", "title": "Emergency Water Conservation Directive and Thermal Permitting Standards for High-Density Computing Facilities (9VAC25-260)", "publisher": "Virginia Department of Environmental Quality & Loudoun County Board of Supervisors", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "hyperscale-pdd-serving-telemetry", "title": "Disaggregated Inference in Production: Decoupling Compute-Bound Prefill from Bandwidth-Bound Decode Across 100,000 GPUs", "publisher": "vLLM and TensorRT-LLM Serving Consortium", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "us-uk-jcps-2026-joint-safety-spec", "title": "Joint Cyber-Physical Safety Evaluation Standard for Dual-Use Frontier Foundation Models (JCPS-2026 / NIST SP 1290-AI)", "publisher": "US AI Safety Institute (NIST) & UK AI Safety Institute (DSIT)", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "arc-anthropic-circuit-breakers-paper", "title": "Representation Circuit Breakers: Real-Time Residual Stream Steering for Autonomous Agent Security", "publisher": "Alignment Research Center & Anthropic Systems Safety Lab", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "swe-bench-verified-diagnostic-audit", "title": "Diagnostic Failure Modes of Autonomous Coding Agents on SWE-bench Verified: The State Synchronization Bottleneck", "publisher": "Open-Source Software Engineering Evaluation Collective", "claimCount": 10, "criticalClaimCount": 10, "numericClaimCount": 9, "claimIds": [ "claim:benchmark.disclosure", "claim:benchmark.items.0", "claim:benchmark.items.1", "claim:benchmark.items.2", "claim:benchmark.items.3", "claim:benchmark.items.4", "claim:benchmark.items.5", "claim:benchmark.items.6", "claim:benchmark.items.7", "claim:dailyDelta.unchanged.0" ], "orphaned": false }, { "sourceId": "openai-delaware-pbc-filing", "title": "Delaware Title 8 § 361 Certificate of Incorporation & California Charitable Trust Reorganization Application", "publisher": "Delaware Division of Corporations & California Office of the Attorney General", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "meta-llama4-frontier-preview-spec", "title": "Llama 4 Frontier Preview Architecture Technical Report: 128-Expert Sparse MoE and Native MXFP4 Training at Scale", "publisher": "Meta AI Research", "claimCount": 1, "criticalClaimCount": 1, "numericClaimCount": 1, "claimIds": [ "claim:models.17" ], "orphaned": false }, { "sourceId": "ocp-mxfp4-microscaling-spec", "title": "Open Compute Project Microscaling Formats (MX) Specification Version 1.0 (MXFP4 / MXFP6 / MXFP8)", "publisher": "Open Compute Project Foundation", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "us-eu-advanced-packaging-accord", "title": "Joint Transatlantic Export Control Accord on Critical Semiconductor Packaging Tools and High-Bandwidth Memory (EAR-2026-4482 / Dual-Use Annex I)", "publisher": "US Department of Commerce BIS & European Commission DG TRADE", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "sk-hynix-samsung-hbm4-contracts", "title": "Quarterly Memory Market Intelligence: Commercial 2027 HBM4 Supply Contracts and Custom Logic Base Allocations", "publisher": "TrendForce & Semiconductor Trade Consortium", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "bair-stanford-quantization-safety-drift", "title": "Asymmetric Representation Degradation: Why Extreme Quantization Erodes Safety Alignment Faster Than General Intelligence", "publisher": "Berkeley AI Research (BAIR) & Stanford Center for Research on Foundation Models (CRFM)", "claimCount": 10, "criticalClaimCount": 10, "numericClaimCount": 9, "claimIds": [ "claim:benchmark.disclosure", "claim:benchmark.items.0", "claim:benchmark.items.1", "claim:benchmark.items.2", "claim:benchmark.items.3", "claim:benchmark.items.4", "claim:benchmark.items.5", "claim:benchmark.items.6", "claim:benchmark.items.7", "claim:dailyDelta.unchanged.0" ], "orphaned": false }, { "sourceId": "google-gemini-2-5-pro-reasoner-paper", "title": "Gemini 2.5 Pro Reasoner Technical Report: Step-Level Supervision, Monte Carlo Tree Search, and Robotics Foundation API", "publisher": "Google DeepMind", "claimCount": 11, "criticalClaimCount": 11, "numericClaimCount": 10, "claimIds": [ "claim:models.15", "claim:benchmark.disclosure", "claim:benchmark.items.0", "claim:benchmark.items.1", "claim:benchmark.items.2", "claim:benchmark.items.3", "claim:benchmark.items.4", "claim:benchmark.items.5", "claim:benchmark.items.6", "claim:benchmark.items.7", "claim:dailyDelta.unchanged.0" ], "orphaned": false }, { "sourceId": "ferc-order-er24-2172-susquehanna", "title": "Federal Energy Regulatory Commission Order Rejecting Amended Interconnection Service Agreement (Docket No. ER24-2172-000)", "publisher": "Federal Energy Regulatory Commission (FERC)", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "mistral-large-3-hybrid-ssm-paper", "title": "Mistral Large 3: Dense 145B Hybrid Selective State-Space and Transformer Model Architecture", "publisher": "Mistral AI Research", "claimCount": 11, "criticalClaimCount": 11, "numericClaimCount": 10, "claimIds": [ "claim:models.16", "claim:benchmark.disclosure", "claim:benchmark.items.0", "claim:benchmark.items.1", "claim:benchmark.items.2", "claim:benchmark.items.3", "claim:benchmark.items.4", "claim:benchmark.items.5", "claim:benchmark.items.6", "claim:benchmark.items.7", "claim:dailyDelta.unchanged.0" ], "orphaned": false }, { "sourceId": "netherlands-duv-export-decree", "title": "Besluit van de Minister voor Buitenlandse Handel inzake de uitvoercontrole van geavanceerde DUV-immersielithografie (Staatscourant 2026-0919)", "publisher": "Ministry of Foreign Trade and Development Cooperation of the Netherlands", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "anthropic-multimodal-prompt-smuggling", "title": "Multi-Modal Boundary Breaches in Autonomous Execution Stacks: Exploiting Visual Representation Layers for Agent Privilege Escalation", "publisher": "Anthropic Alignment & Red-Teaming Research", "claimCount": 10, "criticalClaimCount": 10, "numericClaimCount": 9, "claimIds": [ "claim:benchmark.disclosure", "claim:benchmark.items.0", "claim:benchmark.items.1", "claim:benchmark.items.2", "claim:benchmark.items.3", "claim:benchmark.items.4", "claim:benchmark.items.5", "claim:benchmark.items.6", "claim:benchmark.items.7", "claim:dailyDelta.unchanged.0" ], "orphaned": false }, { "sourceId": "openai-o3-mini-system-card", "title": "OpenAI o3-mini System Card & Technical Documentation: Controllable Test-Time Compute, Mathematical Reasoning, and MicroVM Sandboxing", "publisher": "OpenAI", "claimCount": 4, "criticalClaimCount": 4, "numericClaimCount": 3, "claimIds": [ "claim:models.13", "claim:models.14", "claim:industry.3", "claim:safetyPolicy.11" ], "orphaned": false }, { "sourceId": "tsmc-nvidia-2nm-n2p-agreement", "title": "TSMC Technology Symposium & Commercial Allocation Disclosures: N2P NanoFlex and Backside Power Delivery Network Master Reservations", "publisher": "Taiwan Semiconductor Manufacturing Company (TSMC) / Industry Disclosures", "claimCount": 1, "criticalClaimCount": 1, "numericClaimCount": 1, "claimIds": [ "claim:industry.4" ], "orphaned": false }, { "sourceId": "california-frontier-compute-statute", "title": "California Department of Technology: Enrolled Frontier Artificial Intelligence Compute and Catastrophic Risk Registration Standard (Chapter 842)", "publisher": "California State Government Operations Agency", "claimCount": 2, "criticalClaimCount": 2, "numericClaimCount": 0, "claimIds": [ "claim:industry.5", "claim:safetyPolicy.10" ], "orphaned": false }, { "sourceId": "meta-cmu-agent-collusion-paper", "title": "Sub-Goal Drift and Collusive Oversight in Multi-Agent Delegation Networks (AgentBench-Coop Evaluation)", "publisher": "Meta FAIR / Carnegie Mellon University LTI", "claimCount": 2, "criticalClaimCount": 2, "numericClaimCount": 1, "claimIds": [ "claim:industry.6", "claim:safetyPolicy.12" ], "orphaned": false }, { "sourceId": "jedec-jesd238-hbm4-standard", "title": "JEDEC Solid State Technology Association: JESD238 High Bandwidth Memory 4 (HBM4) Standard Specification", "publisher": "JEDEC Standards Committee", "claimCount": 1, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [ "claim:hardware.1" ], "orphaned": false }, { "sourceId": "eu-ai-office-gpai-scientific-harness", "title": "European AI Office: General Purpose AI Scientific Panel Compliance Test Harness for Frontier Models (>10^25 FLOPs)", "publisher": "European Commission (DG CONNECT)", "claimCount": 2, "criticalClaimCount": 2, "numericClaimCount": 1, "claimIds": [ "claim:industry.1", "claim:safetyPolicy.9" ], "orphaned": false }, { "sourceId": "deepmind-mit-circuit-breakers-paper", "title": "Representation Circuit Breakers for Dual-Use Autonomy: Excising Cyber-Exploit Vectors from Transformer Residual Streams Without Utility Loss", "publisher": "Google DeepMind / MIT CSAIL", "claimCount": 1, "criticalClaimCount": 1, "numericClaimCount": 1, "claimIds": [ "claim:safetyPolicy.8" ], "orphaned": false }, { "sourceId": "constellation-microsoft-crane-clean-energy-pjm", "title": "PJM Interconnection & Constellation Energy: System Impact Study and Interconnection Agreement for Crane Clean Energy Center (835MW Nuclear Restart)", "publisher": "PJM Interconnection / SEC Form 8-K", "claimCount": 1, "criticalClaimCount": 1, "numericClaimCount": 1, "claimIds": [ "claim:industry.2" ], "orphaned": false }, { "sourceId": "openai-o3-pro-system-card", "title": "OpenAI o3-pro System Card: Frontier Reasoning, Test-Time Tree Search with Native Multi-Step Tool Execution, and Autonomous Coding Evaluations", "publisher": "OpenAI", "claimCount": 39, "criticalClaimCount": 39, "numericClaimCount": 34, "claimIds": [ "claim:snapshots.0", "claim:models.9", "claim:apiPriceComparison.0", "claim:benchmark.disclosure", "claim:benchmark.items.0", "claim:benchmark.items.1", "claim:benchmark.items.2", "claim:benchmark.items.3", "claim:benchmark.items.4", "claim:benchmark.items.5", "claim:benchmark.items.6", "claim:benchmark.items.7", "claim:benchmarkDimensions.0.status", "claim:benchmarkDimensions.0.items.0", "claim:benchmarkDimensions.0.items.1", "claim:benchmarkDimensions.0.items.2", "claim:benchmarkDimensions.0.items.3", "claim:benchmarkDimensions.0.items.4", "claim:benchmarkDimensions.0.items.5", "claim:benchmarkDimensions.0.items.6", "claim:benchmarkDimensions.0.items.7", "claim:benchmarkDimensions.1.status", "claim:benchmarkDimensions.1.items.0", "claim:benchmarkDimensions.1.items.1", "claim:benchmarkDimensions.1.items.2", "claim:benchmarkDimensions.1.items.3", "claim:benchmarkDimensions.1.items.4", "claim:benchmarkDimensions.1.items.5", "claim:benchmarkDimensions.1.items.6", "claim:benchmarkDimensions.1.items.7", "claim:benchmarkDimensions.6.status", "claim:benchmarkDimensions.6.items.0", "claim:benchmarkDimensions.6.items.1", "claim:benchmarkDimensions.6.items.2", "claim:benchmarkDimensions.6.items.3", "claim:benchmarkDimensions.6.items.4", "claim:benchmarkDimensions.6.items.5", "claim:benchmarkDimensions.6.items.6", "claim:dailyDelta.unchanged.0" ], "orphaned": false }, { "sourceId": "tsmc-n2p-gaa-bspdn-announcement", "title": "TSMC Completes Tape-Out of 2nm (N2P) GAA Silicon with Backside Power Delivery (Super Power Rail) for Next-Generation AI Accelerators", "publisher": "TSMC / NVIDIA Press Center", "claimCount": 2, "criticalClaimCount": 1, "numericClaimCount": 1, "claimIds": [ "claim:snapshots.1", "claim:hardware.0" ], "orphaned": false }, { "sourceId": "alibaba-qwen-2-5-coder-reasoner-paper", "title": "Qwen 2.5-Coder-32B-Reasoner: Open-Weight Test-Time Search Code Agent with Step-Level Process Reward Supervision", "publisher": "Alibaba Cloud & Qwen Team", "claimCount": 22, "criticalClaimCount": 22, "numericClaimCount": 20, "claimIds": [ "claim:snapshots.2", "claim:models.10", "claim:apiPriceComparison.1", "claim:benchmark.disclosure", "claim:benchmark.items.0", "claim:benchmark.items.1", "claim:benchmark.items.2", "claim:benchmark.items.3", "claim:benchmark.items.4", "claim:benchmark.items.5", "claim:benchmark.items.6", "claim:benchmark.items.7", "claim:benchmarkDimensions.1.status", "claim:benchmarkDimensions.1.items.0", "claim:benchmarkDimensions.1.items.1", "claim:benchmarkDimensions.1.items.2", "claim:benchmarkDimensions.1.items.3", "claim:benchmarkDimensions.1.items.4", "claim:benchmarkDimensions.1.items.5", "claim:benchmarkDimensions.1.items.6", "claim:benchmarkDimensions.1.items.7", "claim:dailyDelta.unchanged.0" ], "orphaned": false }, { "sourceId": "ferc-doe-clean-energy-ai-interconnect-nopr", "title": "Federal Energy Regulatory Commission & Department of Energy: Emergency NOPR on Fast-Track Clean Energy Interconnections for Large-Scale AI Data Centers (Docket No. RM26-11-000)", "publisher": "Federal Energy Regulatory Commission (FERC) / US Department of Energy", "claimCount": 1, "criticalClaimCount": 1, "numericClaimCount": 1, "claimIds": [ "claim:industry.0" ], "orphaned": false }, { "sourceId": "aisi-joint-multimodal-agent-sandbox-audit", "title": "US and UK AI Safety Institutes: Joint Empirical Audit on Multimodal Visual Prompt Smuggling and Mandatory Sandbox Virtualization for Desktop Agents", "publisher": "US AI Safety Institute (NIST) / UK AI Safety Institute (DSIT)", "claimCount": 12, "criticalClaimCount": 12, "numericClaimCount": 11, "claimIds": [ "claim:snapshots.3", "claim:safetyPolicy.7", "claim:benchmark.disclosure", "claim:benchmark.items.0", "claim:benchmark.items.1", "claim:benchmark.items.2", "claim:benchmark.items.3", "claim:benchmark.items.4", "claim:benchmark.items.5", "claim:benchmark.items.6", "claim:benchmark.items.7", "claim:dailyDelta.unchanged.0" ], "orphaned": false }, { "sourceId": "anthropic-claude-3-7-sonnet-system-card", "title": "Anthropic Claude 3.7 Sonnet: Hybrid Reasoning Architecture, Extended Thinking Token Scaling, and Sub-Pixel Computer Use v2", "publisher": "Anthropic", "claimCount": 40, "criticalClaimCount": 40, "numericClaimCount": 35, "claimIds": [ "claim:models.8", "claim:models.11", "claim:models.12", "claim:apiPriceComparison.2", "claim:benchmark.disclosure", "claim:benchmark.items.0", "claim:benchmark.items.1", "claim:benchmark.items.2", "claim:benchmark.items.3", "claim:benchmark.items.4", "claim:benchmark.items.5", "claim:benchmark.items.6", "claim:benchmark.items.7", "claim:benchmarkDimensions.0.status", "claim:benchmarkDimensions.0.items.0", "claim:benchmarkDimensions.0.items.1", "claim:benchmarkDimensions.0.items.2", "claim:benchmarkDimensions.0.items.3", "claim:benchmarkDimensions.0.items.4", "claim:benchmarkDimensions.0.items.5", "claim:benchmarkDimensions.0.items.6", "claim:benchmarkDimensions.0.items.7", "claim:benchmarkDimensions.1.status", "claim:benchmarkDimensions.1.items.0", "claim:benchmarkDimensions.1.items.1", "claim:benchmarkDimensions.1.items.2", "claim:benchmarkDimensions.1.items.3", "claim:benchmarkDimensions.1.items.4", "claim:benchmarkDimensions.1.items.5", "claim:benchmarkDimensions.1.items.6", "claim:benchmarkDimensions.1.items.7", "claim:benchmarkDimensions.6.status", "claim:benchmarkDimensions.6.items.0", "claim:benchmarkDimensions.6.items.1", "claim:benchmarkDimensions.6.items.2", "claim:benchmarkDimensions.6.items.3", "claim:benchmarkDimensions.6.items.4", "claim:benchmarkDimensions.6.items.5", "claim:benchmarkDimensions.6.items.6", "claim:dailyDelta.unchanged.0" ], "orphaned": false }, { "sourceId": "google-workspace-deep-research-announcement", "title": "Google Workspace Updates: Autonomous Multimodal Deep Research Integration in Google Docs and Gemini Advanced", "publisher": "Google Workspace / Google Cloud", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "eu-ai-office-gpai-systemic-risk-code", "title": "European AI Office: Finalized Draft Code of Practice and Compliance Metrics for General-Purpose AI Models with Systemic Risk (>10^25 FLOPs)", "publisher": "European Commission / European AI Office", "claimCount": 1, "criticalClaimCount": 1, "numericClaimCount": 1, "claimIds": [ "claim:safetyPolicy.5" ], "orphaned": false }, { "sourceId": "cmu-uk-aisi-agent-injection-preprint", "title": "Indirect Prompt Injection via Multimodal Environment Interaction in Frontier GUI Agents and Mandatory MicroVM Sandboxing", "publisher": "Carnegie Mellon University CyLab / UK AI Safety Institute", "claimCount": 1, "criticalClaimCount": 1, "numericClaimCount": 1, "claimIds": [ "claim:safetyPolicy.6" ], "orphaned": false }, { "sourceId": "swe-bench-verified-october-audit", "title": "SWE-bench Verified & Terminal-Bench October 2026 Frontier Coding Leaderboard: Test-Time Reasoning vs. Open-Weight Architectures", "publisher": "CodeEval Foundation / SWE-bench Consortium", "claimCount": 19, "criticalClaimCount": 19, "numericClaimCount": 17, "claimIds": [ "claim:benchmark.disclosure", "claim:benchmark.items.0", "claim:benchmark.items.1", "claim:benchmark.items.2", "claim:benchmark.items.3", "claim:benchmark.items.4", "claim:benchmark.items.5", "claim:benchmark.items.6", "claim:benchmark.items.7", "claim:benchmarkDimensions.1.status", "claim:benchmarkDimensions.1.items.0", "claim:benchmarkDimensions.1.items.1", "claim:benchmarkDimensions.1.items.2", "claim:benchmarkDimensions.1.items.3", "claim:benchmarkDimensions.1.items.4", "claim:benchmarkDimensions.1.items.5", "claim:benchmarkDimensions.1.items.6", "claim:benchmarkDimensions.1.items.7", "claim:dailyDelta.unchanged.0" ], "orphaned": false }, { "sourceId": "aws-trainium-3-custom-silicon-pricing", "title": "Amazon EC2 Trn3 Instances General Availability: Scaling Frontier Inference Workloads at 50% Lower Cost per Accelerator-Hour", "publisher": "Amazon Web Services", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "openai-operator-sandbox-architecture", "title": "OpenAI Operator Architecture: MicroVM Hardware Isolation, Permission Gating, and Dual-Loop Verification for Web Agents", "publisher": "OpenAI", "claimCount": 20, "criticalClaimCount": 20, "numericClaimCount": 17, "claimIds": [ "claim:models.5", "claim:safetyPolicy.4", "claim:benchmark.disclosure", "claim:benchmark.items.0", "claim:benchmark.items.1", "claim:benchmark.items.2", "claim:benchmark.items.3", "claim:benchmark.items.4", "claim:benchmark.items.5", "claim:benchmark.items.6", "claim:benchmark.items.7", "claim:benchmarkDimensions.6.status", "claim:benchmarkDimensions.6.items.0", "claim:benchmarkDimensions.6.items.1", "claim:benchmarkDimensions.6.items.2", "claim:benchmarkDimensions.6.items.3", "claim:benchmarkDimensions.6.items.4", "claim:benchmarkDimensions.6.items.5", "claim:benchmarkDimensions.6.items.6", "claim:dailyDelta.unchanged.0" ], "orphaned": false }, { "sourceId": "meta-llama-4-speculative-decoding-preview", "title": "Meta AI Engineering: Llama 4 Architectural Blueprint, Native Speculative Decoding, and MoE Routing at Scale", "publisher": "Meta AI / Fundamental AI Research", "claimCount": 11, "criticalClaimCount": 11, "numericClaimCount": 10, "claimIds": [ "claim:models.7", "claim:benchmark.disclosure", "claim:benchmark.items.0", "claim:benchmark.items.1", "claim:benchmark.items.2", "claim:benchmark.items.3", "claim:benchmark.items.4", "claim:benchmark.items.5", "claim:benchmark.items.6", "claim:benchmark.items.7", "claim:dailyDelta.unchanged.0" ], "orphaned": false }, { "sourceId": "nvidia-tensorrt-llm-fp4-blackwell", "title": "NVIDIA TensorRT-LLM 0.16 General Availability: FP4 Microscaling Acceleration on Blackwell GB200 Systems", "publisher": "NVIDIA Technical Blog", "claimCount": 10, "criticalClaimCount": 10, "numericClaimCount": 9, "claimIds": [ "claim:benchmark.disclosure", "claim:benchmark.items.0", "claim:benchmark.items.1", "claim:benchmark.items.2", "claim:benchmark.items.3", "claim:benchmark.items.4", "claim:benchmark.items.5", "claim:benchmark.items.6", "claim:benchmark.items.7", "claim:dailyDelta.unchanged.0" ], "orphaned": false }, { "sourceId": "nist-aisi-agentic-privilege-escalation-bench", "title": "US Artificial Intelligence Safety Institute (NIST AISI): Standardized Test Suite for Agentic Privilege Escalation and MicroVM Containment", "publisher": "National Institute of Standards and Technology (NIST)", "claimCount": 19, "criticalClaimCount": 19, "numericClaimCount": 16, "claimIds": [ "claim:safetyPolicy.3", "claim:benchmark.disclosure", "claim:benchmark.items.0", "claim:benchmark.items.1", "claim:benchmark.items.2", "claim:benchmark.items.3", "claim:benchmark.items.4", "claim:benchmark.items.5", "claim:benchmark.items.6", "claim:benchmark.items.7", "claim:benchmarkDimensions.6.status", "claim:benchmarkDimensions.6.items.0", "claim:benchmarkDimensions.6.items.1", "claim:benchmarkDimensions.6.items.2", "claim:benchmarkDimensions.6.items.3", "claim:benchmarkDimensions.6.items.4", "claim:benchmarkDimensions.6.items.5", "claim:benchmarkDimensions.6.items.6", "claim:dailyDelta.unchanged.0" ], "orphaned": false }, { "sourceId": "google-deepmind-gemini-2-5-thinking-vertex", "title": "Google Cloud Vertex AI: Gemini 2.5 Flash and Pro Thinking Mode with Native Recursive Code Execution", "publisher": "Google Cloud / Google DeepMind", "claimCount": 1, "criticalClaimCount": 1, "numericClaimCount": 1, "claimIds": [ "claim:models.6" ], "orphaned": false }, { "sourceId": "anthropic-claude-3-8-system-card", "title": "Anthropic Claude 3.8 Architecture & System Card: Computer Use v2.5, POSIX Bash Sandboxing, and SWE-bench Verified Benchmark", "publisher": "Anthropic PBC", "claimCount": 18, "criticalClaimCount": 18, "numericClaimCount": 16, "claimIds": [ "claim:benchmark.disclosure", "claim:benchmark.items.0", "claim:benchmark.items.1", "claim:benchmark.items.2", "claim:benchmark.items.3", "claim:benchmark.items.4", "claim:benchmark.items.5", "claim:benchmark.items.6", "claim:benchmark.items.7", "claim:benchmarkDimensions.6.status", "claim:benchmarkDimensions.6.items.0", "claim:benchmarkDimensions.6.items.1", "claim:benchmarkDimensions.6.items.2", "claim:benchmarkDimensions.6.items.3", "claim:benchmarkDimensions.6.items.4", "claim:benchmarkDimensions.6.items.5", "claim:benchmarkDimensions.6.items.6", "claim:dailyDelta.unchanged.0" ], "orphaned": false }, { "sourceId": "deepseek-v3-5-architecture-report", "title": "DeepSeek-V3.5 Technical Report: Multi-Head Latent Attention (MLA) Scaling and DualPipe Parallelism across 2,048 Compute Nodes", "publisher": "DeepSeek AI", "claimCount": 18, "criticalClaimCount": 18, "numericClaimCount": 16, "claimIds": [ "claim:benchmark.disclosure", "claim:benchmark.items.0", "claim:benchmark.items.1", "claim:benchmark.items.2", "claim:benchmark.items.3", "claim:benchmark.items.4", "claim:benchmark.items.5", "claim:benchmark.items.6", "claim:benchmark.items.7", "claim:benchmarkDimensions.6.status", "claim:benchmarkDimensions.6.items.0", "claim:benchmarkDimensions.6.items.1", "claim:benchmarkDimensions.6.items.2", "claim:benchmarkDimensions.6.items.3", "claim:benchmarkDimensions.6.items.4", "claim:benchmarkDimensions.6.items.5", "claim:benchmarkDimensions.6.items.6", "claim:dailyDelta.unchanged.0" ], "orphaned": false }, { "sourceId": "tsmc-n2p-backside-power-disclosure", "title": "TSMC N2P Risk Production: Backside Power Delivery Network (BSPDN) and NanoFlex Transistor Integration for 2027 AI Accelerators", "publisher": "Taiwan Semiconductor Manufacturing Company (TSMC)", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "nvidia-blackwell-ultra-nvl72-fp4", "title": "NVIDIA Blackwell Ultra GB200 NVL72 System Architecture: 3.2x Multi-Turn Chain-of-Thought Inference via FP4 Microscaling", "publisher": "NVIDIA Technical Blog", "claimCount": 10, "criticalClaimCount": 10, "numericClaimCount": 9, "claimIds": [ "claim:benchmark.disclosure", "claim:benchmark.items.0", "claim:benchmark.items.1", "claim:benchmark.items.2", "claim:benchmark.items.3", "claim:benchmark.items.4", "claim:benchmark.items.5", "claim:benchmark.items.6", "claim:benchmark.items.7", "claim:dailyDelta.unchanged.0" ], "orphaned": false }, { "sourceId": "nist-aisi-c2pa-watermarking-guidance", "title": "US Artificial Intelligence Safety Institute (NIST AISI): Federal Guidance on C2PA Cryptographic Provenance Containers for Generative APIs", "publisher": "National Institute of Standards and Technology (NIST)", "claimCount": 1, "criticalClaimCount": 1, "numericClaimCount": 0, "claimIds": [ "claim:safetyPolicy.2" ], "orphaned": false }, { "sourceId": "vllm-speculative-decoding-llama4", "title": "vLLM 0.7.2 Release: Zero-Overhead Speculative Decoding Kernels for Llama 4 and DeepSeek MLA Architectures", "publisher": "vLLM Project / UC Berkeley LMSYS", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "cerebras-cs3-wafer-scale-cluster", "title": "Cerebras CS-3 10,000-Chip Inference Cluster Deployment: 1,820 tokens/sec Sustained Throughput for 70B Models", "publisher": "Cerebras Systems", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "openai-chatgpt-intelligent-ui-gpt6", "title": "OpenAI Intelligent UI Rollout & GPT-6 General Availability Deployment Whitepaper", "publisher": "OpenAI", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "anthropic-claude-motion-usage-policy", "title": "Anthropic Claude Motion Engine, Dashboards Beta, and System Call Containment Policy", "publisher": "Anthropic PBC", "claimCount": 19, "criticalClaimCount": 19, "numericClaimCount": 16, "claimIds": [ "claim:safetyPolicy.1", "claim:benchmark.disclosure", "claim:benchmark.items.0", "claim:benchmark.items.1", "claim:benchmark.items.2", "claim:benchmark.items.3", "claim:benchmark.items.4", "claim:benchmark.items.5", "claim:benchmark.items.6", "claim:benchmark.items.7", "claim:benchmarkDimensions.6.status", "claim:benchmarkDimensions.6.items.0", "claim:benchmarkDimensions.6.items.1", "claim:benchmarkDimensions.6.items.2", "claim:benchmarkDimensions.6.items.3", "claim:benchmarkDimensions.6.items.4", "claim:benchmarkDimensions.6.items.5", "claim:benchmarkDimensions.6.items.6", "claim:dailyDelta.unchanged.0" ], "orphaned": false }, { "sourceId": "qwen-image-2-1-turbo-announcement", "title": "Qwen-Image-2.1-Turbo Technical Specification: 8-Step Euler Flow-Matching with 64-Channel RGBA VAE", "publisher": "Alibaba Tongyi Lab", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "google-synthid-universal-verification", "title": "Google DeepMind SynthID Universal Detection Protocol & Multi-Vendor Provenance Bridge", "publisher": "Google DeepMind", "claimCount": 1, "criticalClaimCount": 1, "numericClaimCount": 0, "claimIds": [ "claim:safetyPolicy.0" ], "orphaned": false }, { "sourceId": "deepseek-pre-ipo-ascend-cluster", "title": "DeepSeek Infrastructure Architecture: 160,000-Accelerator Huawei Ascend CANN Cluster & Capital Disclosures", "publisher": "DeepSeek AI / Industry Regulatory Filings", "claimCount": 1, "criticalClaimCount": 1, "numericClaimCount": 1, "claimIds": [ "claim:models.4" ], "orphaned": false }, { "sourceId": "hyperscaler-capex-tsmc-expansion", "title": "Q3 2026 Hyperscale Infrastructure Consensus & TSMC Arizona $100B Additional Phase Allocation", "publisher": "TSMC Investor Relations / Financial Disclosures", "claimCount": 0, "criticalClaimCount": 0, "numericClaimCount": 0, "claimIds": [], "orphaned": true }, { "sourceId": "meta-muse-windows-native", "title": "Meta Muse Desktop Architecture: Native Windows Runtime, Local Memory Buffers, and Cross-Platform Integration", "publisher": "Meta Platforms & Microsoft", "claimCount": 18, "criticalClaimCount": 18, "numericClaimCount": 16, "claimIds": [ "claim:benchmark.disclosure", "claim:benchmark.items.0", "claim:benchmark.items.1", "claim:benchmark.items.2", "claim:benchmark.items.3", "claim:benchmark.items.4", "claim:benchmark.items.5", "claim:benchmark.items.6", "claim:benchmark.items.7", "claim:benchmarkDimensions.6.status", "claim:benchmarkDimensions.6.items.0", "claim:benchmarkDimensions.6.items.1", "claim:benchmarkDimensions.6.items.2", "claim:benchmarkDimensions.6.items.3", "claim:benchmarkDimensions.6.items.4", "claim:benchmarkDimensions.6.items.5", "claim:benchmarkDimensions.6.items.6", "claim:dailyDelta.unchanged.0" ], "orphaned": false }, { "sourceId": "openai-gpt6-ga-whitepaper", "title": "OpenAI GPT-6 (Sol & Luna) Dual-Engine Production Deployment Specification", "publisher": "OpenAI", "claimCount": 11, "criticalClaimCount": 10, "numericClaimCount": 5, "claimIds": [ "claim:headline", "claim:dek", "claim:plainEnglish", "claim:whatChanged.0", "claim:mattersToday.0", "claim:executiveBriefing.0.summary", "claim:executiveBriefing.0.whatHappened", "claim:executiveBriefing.0.whyItMatters", "claim:executiveBriefing.0.action", "claim:models.1", "claim:models.3" ], "orphaned": false }, { "sourceId": "tsmc-arizona-phase23-expansion", "title": "TSMC Board Authorization of $100B Multi-Phase Phoenix Arizona Foundry Expansion", "publisher": "TSMC Investor Relations / SEC Form 6-K", "claimCount": 10, "criticalClaimCount": 8, "numericClaimCount": 3, "claimIds": [ "claim:headline", "claim:dek", "claim:plainEnglish", "claim:whatChanged.1", "claim:mattersToday.1", "claim:watchNext.1", "claim:executiveBriefing.1.summary", "claim:executiveBriefing.1.whatHappened", "claim:executiveBriefing.1.whyItMatters", "claim:executiveBriefing.1.action" ], "orphaned": false }, { "sourceId": "anthropic-claude-motion-posix-policy", "title": "Anthropic Claude Motion Engine GA and Mandatory Host POSIX MicroVM Containment Policy", "publisher": "Anthropic PBC", "claimCount": 11, "criticalClaimCount": 9, "numericClaimCount": 2, "claimIds": [ "claim:headline", "claim:dek", "claim:plainEnglish", "claim:whatChanged.2", "claim:mattersToday.2", "claim:watchNext.2", "claim:executiveBriefing.2.summary", "claim:executiveBriefing.2.whatHappened", "claim:executiveBriefing.2.whyItMatters", "claim:executiveBriefing.2.action", "claim:models.2" ], "orphaned": false }, { "sourceId": "deepseek-ascend-cann9-cluster-whitepaper", "title": "DeepSeek 160,000-Accelerator Megacluster on Huawei Ascend CANN 9.0 Architecture", "publisher": "DeepSeek AI / Industry Filings", "claimCount": 12, "criticalClaimCount": 12, "numericClaimCount": 9, "claimIds": [ "claim:plainEnglish", "claim:whatChanged.3", "claim:benchmark.disclosure", "claim:benchmark.items.0", "claim:benchmark.items.1", "claim:benchmark.items.2", "claim:benchmark.items.3", "claim:benchmark.items.4", "claim:benchmark.items.5", "claim:benchmark.items.6", "claim:benchmark.items.7", "claim:dailyDelta.unchanged.0" ], "orphaned": false }, { "sourceId": "c2pa-synthid-universal-resolution", "title": "C2PA Ratification of Google DeepMind SynthID Universal Detection Standard", "publisher": "C2PA & Google DeepMind", "claimCount": 4, "criticalClaimCount": 3, "numericClaimCount": 1, "claimIds": [ "claim:dek", "claim:plainEnglish", "claim:whatChanged.4", "claim:watchNext.0" ], "orphaned": false }, { "sourceId": "google-gemini-38-multimodal-release", "title": "Google DeepMind Gemini 3.8 (Flash & Pro) Architecture Whitepaper & System API Documentation", "publisher": "Google DeepMind", "claimCount": 1, "criticalClaimCount": 1, "numericClaimCount": 1, "claimIds": [ "claim:models.0" ], "orphaned": false } ], "blockers": [ { "code": "weak-claim-evidence", "claimId": "claim:models.18", "path": "models.18", "message": "models.18 is supported only by grade D/E evidence." }, { "code": "weak-claim-evidence", "claimId": "claim:markets.0", "path": "markets.0", "message": "markets.0 is supported only by grade D/E evidence." }, { "code": "weak-claim-evidence", "claimId": "claim:markets.1", "path": "markets.1", "message": "markets.1 is supported only by grade D/E evidence." }, { "code": "weak-claim-evidence", "claimId": "claim:markets.2", "path": "markets.2", "message": "markets.2 is supported only by grade D/E evidence." }, { "code": "critical-coverage-below-threshold", "message": "Critical claim citation coverage is 97%; 100% is required." }, { "code": "numeric-coverage-below-threshold", "message": "Numerical claim citation coverage is 97%; 100% is required." } ], "warnings": [], "publication": { "status": "blocked", "blockerCount": 6, "warningCount": 0, "requiredAcknowledgements": [ "claimsReviewed", "citationCoverageReviewed", "numericalClaimsReviewed" ], "rule": "Every critical and numerical claim must have at least one valid explicit source. Invalid source references and critical claims supported only by grade D/E evidence block publication." }, "methodology": { "claimIdentity": "Stable report-object path", "citationResolutionOrder": [ "record-level sourceId/sourceIds", "parent benchmark sourceIds", "explicit claimEvidence map" ], "criticalCoverageThreshold": 100, "numericCoverageThreshold": 100, "sourceOnlyTextOrMetadataChangesDoNotAlterClaimIdentity": true, "claimEvidenceMapSupported": true } }, "revisionLedger": { "version": "1.0", "editionDate": "2026-10-19", "generatedAt": "2026-10-19T07:00:00-04:00", "baselineEditionDate": null, "baselineAvailable": false, "entryCount": 0, "counts": { "total": 0, "corrections": 0, "clarifications": 0, "retractions": 0, "supersessions": 0, "draft": 0, "approved": 0, "published": 0, "critical": 0, "material": 0, "minor": 0 }, "entries": [], "chainHead": null, "changedPublishedClaimPaths": [], "silentlyChangedClaimPaths": [], "blockers": [], "warnings": [], "publication": { "status": "publishable", "blockerCount": 0, "warningCount": 0, "requiredAcknowledgements": [ "correctionsReviewed", "beforeAfterReviewed", "affectedPagesReviewed", "revisionChainReviewed" ], "rule": "Material changes to an already published edition require an approved, evidence-backed, hash-chained revision entry. Silent changes and incomplete retractions block publication." }, "methodology": { "immutableBeforeAfterHashes": true, "chainedRevisionHashes": true, "silentPublishedChangesBlocked": true, "allowedTypes": [ "correction", "clarification", "retraction", "supersession" ], "allowedStatuses": [ "draft", "approved", "published" ], "allowedSeverities": [ "critical", "material", "minor" ] } }, "methodologyVersions": { "historicalDelta": 1, "trendTimeline": 1, "pricingHistory": 1, "evidenceQuality": 1, "claimProvenance": 1, "correctionLedger": 1 } }