Metatron Intelligence
Today: the five AI headlines of the day → and the AI Wiki
What an entity record is How to read this page

A named actor, followed across desks

An organisation, model, chip or programme, with the statements attached to it, so that the same actor can be tracked without re-reading every domain.

Statements carry their own sources

Each statement on this page is bound to a source and a date. The entity page collects them; it does not average them into a verdict.

Absence is not evidence

An entity with few statements has been observed less, not necessarily done less. Read the count as coverage, never as significance.

New to this publication?

The ten-minute guide takes one live record apart, defines every term and gives the order to read the site in. Start here →

74 curated records · 383 statements · 38 qualified relations · Atlas r0056

Entity Intelligence

Stable identities before comparison: provider, organization, legal entity, team, service, family, release, alias and benchmark protocol travel with source, scope and boundary.

Every identity is reviewed, never guessed.

The registry keeps provider brands, organizations, legal entities, labs or teams, provider services, model families, dated releases, deployment aliases and benchmark definitions disjoint. Benchmark passports bind task, metric, evaluation conditions and comparability limits before any result can be interpreted. Exact source-URL routing is not a subject inference, and a reviewed identity is not a legal determination. Current status remains unknown unless separately time-qualified. Remaining explicit coverage gaps: legal expression, artifact or distribution, deployment product, country, jurisdiction, incident.

Inspect registry JSON →Inspect record schema →Inspect statement schema →Inspect registry schema →Inspect coverage gaps →Download versioned dataset →
  1. model family · Command R+Command R+0 exact claim connections · 4 source-qualified statements
  2. dated model version · gpt-4o-2024-05-13GPT-4o snapshot 2024-05-130 exact claim connections · 5 source-qualified statements
  3. provider · meta.comMeta0 exact claim connections · 4 source-qualified statements
  4. deployment alias · mistral-medium-latestmistral-medium-latest0 exact claim connections · 4 source-qualified statements
  5. model family · Llama 3.1Llama 3.10 exact claim connections · 4 source-qualified statements
  6. governance document · 5.2.0CVE Record Format0 exact claim connections · 6 source-qualified statements
  7. legal entity · Hangzhou DeepSeek Artificial Intelligence Co., Ltd.Hangzhou DeepSeek Artificial Intelligence Co., Ltd.0 exact claim connections · 4 source-qualified statements
  8. organization · Google operational organizationGoogle0 exact claim connections · 4 source-qualified statements
  9. provider service · claude_amazon_bedrock_channelClaude on Amazon Bedrock0 exact claim connections · 4 source-qualified statements
  10. provider service · google_gemini_apiGemini API0 exact claim connections · 4 source-qualified statements
  11. model family · GPT-4oGPT-4o2 exact claim connections · 4 source-qualified statements
  12. provider service · deepseek_hosted_apiDeepSeek API0 exact claim connections · 4 source-qualified statements
  13. benchmark · google/BIG-bench@092b196c1f8f14a54bbc62f24759d43bde46dd3bBIG-bench0 exact claim connections · 11 source-qualified statements
  14. lab or team · Cohere LabsCohere Labs0 exact claim connections · 4 source-qualified statements
  15. legal entity · Anthropic, PBCAnthropic, PBC0 exact claim connections · 4 source-qualified statements
  16. legal resource · Regulation (EU) 2024/1689Regulation (EU) 2024/1689 laying down harmonised rules on artificial intelligence0 exact claim connections · 6 source-qualified statements
  17. benchmark · arena-leaderboard-policy-2026-09-01Arena Leaderboard Policy0 exact claim connections · 11 source-qualified statements
  18. dated model version · grok-4-0709Grok 4 snapshot 07090 exact claim connections · 5 source-qualified statements
  19. benchmark · openai/human-eval@6d43fb980f9fee3c892a914eda09951f772ad10dHumanEval0 exact claim connections · 11 source-qualified statements
  20. legal entity · Cohere Inc.Cohere Inc.0 exact claim connections · 4 source-qualified statements
  21. provider service · anthropic_direct_apiAnthropic API0 exact claim connections · 4 source-qualified statements
  22. model family · Command ACommand A0 exact claim connections · 4 source-qualified statements
  23. organization · Cohere operational organizationCohere0 exact claim connections · 4 source-qualified statements
  24. governance document · 1.0Artificial Intelligence Risk Management Framework: Generative Artificial Intelligence Profile2 exact claim connections · 6 source-qualified statements
  25. dated model version · mistral-small-2503Mistral Small 3.10 exact claim connections · 5 source-qualified statements
  26. deployment alias · mistral-medium-3mistral-medium-30 exact claim connections · 4 source-qualified statements
  27. provider · x.aixAI0 exact claim connections · 4 source-qualified statements
  28. model family · DeepSeek-R1DeepSeek-R10 exact claim connections · 4 source-qualified statements
  29. governance document · 2025-02-28Towards a common reporting framework for AI incidents0 exact claim connections · 6 source-qualified statements
  30. dated model version · claude-3-5-sonnet-20240620Claude 3.5 Sonnet 202406200 exact claim connections · 5 source-qualified statements
  31. provider service · claude_google_vertex_channelClaude on Google Vertex AI0 exact claim connections · 4 source-qualified statements
  32. model family · Gemini 2.5 ProGemini 2.5 Pro0 exact claim connections · 4 source-qualified statements
  33. dated model version · command-r-plus-04-2024Command R+ 04-20240 exact claim connections · 4 source-qualified statements
  34. provider service · openai_apiOpenAI API0 exact claim connections · 4 source-qualified statements
  35. provider · mistral.aiMistral AI0 exact claim connections · 4 source-qualified statements
  36. provider service · google_cloud_vertex_aiVertex AI0 exact claim connections · 4 source-qualified statements
  37. dated model version · command-a-plus-05-2026Command A+0 exact claim connections · 5 source-qualified statements
  38. provider · openai.comOpenAI0 exact claim connections · 4 source-qualified statements
  39. organization · Meta Platforms, Inc. operating organizationMeta Platforms, Inc.0 exact claim connections · 4 source-qualified statements
  40. provider · anthropic.comAnthropic0 exact claim connections · 4 source-qualified statements
  41. provider service · llama_api_preview_2025_04_29Llama API limited free preview0 exact claim connections · 4 source-qualified statements
  42. deployment alias · command-r-pluscommand-r-plus0 exact claim connections · 4 source-qualified statements
  43. provider service · meta_model_api_public_previewMeta Model API0 exact claim connections · 4 source-qualified statements
  44. legal entity · Anthropic Ireland, LimitedAnthropic Ireland, Limited0 exact claim connections · 4 source-qualified statements
  45. governance document · BOD-20-01BOD 20-01: Develop and Publish a Vulnerability Disclosure Policy0 exact claim connections · 6 source-qualified statements
  46. governance document · 2023ISO/IEC 42001:2023 — Artificial intelligence management system0 exact claim connections · 6 source-qualified statements
  47. benchmark · idavidrein/gpqa@56686c06f5e19865c153de0fdb11be3890014df7GPQA0 exact claim connections · 11 source-qualified statements
  48. model family · Mistral SmallMistral Small0 exact claim connections · 4 source-qualified statements
  49. model family · Claude 3.5 SonnetClaude 3.5 Sonnet0 exact claim connections · 4 source-qualified statements
  50. provider service · spacexai_enterprise_apiSpaceXAI API0 exact claim connections · 4 source-qualified statements
  51. benchmark · MLPerf Inference 6.0MLPerf Inference v6.00 exact claim connections · 11 source-qualified statements
  52. dated model version · Llama-3.1-405B-InstructLlama 3.1 405B Instruct0 exact claim connections · 5 source-qualified statements
  53. provider service · cohere_direct_api_saasCohere API SaaS Services0 exact claim connections · 4 source-qualified statements
  54. benchmark · SWE-bench_VerifiedSWE-bench Verified0 exact claim connections · 11 source-qualified statements
  55. organization · Anthropic operational organizationAnthropic0 exact claim connections · 4 source-qualified statements
  56. legal entity · Mistral AI, French limited joint-stock corporation, Paris registration 952 418 325Mistral AI (Paris registration 952 418 325)0 exact claim connections · 4 source-qualified statements
  57. dated model version · DeepSeek-R1-0528DeepSeek-R1-05280 exact claim connections · 4 source-qualified statements
  58. benchmark · hendrycks/test@4450500f923c49f1fb1dd3d99108a0bd9717b660Massive Multitask Language Understanding (MMLU)0 exact claim connections · 11 source-qualified statements
  59. deployment alias · gpt-4ogpt-4o0 exact claim connections · 4 source-qualified statements
  60. lab or team · Google DeepMind teamGoogle DeepMind0 exact claim connections · 4 source-qualified statements
  61. dated model version · DeepSeek-R1DeepSeek-R1 first-party weights release0 exact claim connections · 5 source-qualified statements
  62. provider · deepseek.comDeepSeek0 exact claim connections · 4 source-qualified statements
  63. model family · GrokGrok0 exact claim connections · 4 source-qualified statements
  64. benchmark · METR Time Horizon 1.1METR Task-Completion Time Horizon 1.11 exact claim connections · 11 source-qualified statements
  65. provider service · mistral_ai_infrastructure_apiMistral AI API0 exact claim connections · 4 source-qualified statements
  66. dated model version · gemini-2.5-pro-preview-06-05Gemini 2.5 Pro preview 06-050 exact claim connections · 5 source-qualified statements
  67. deployment alias · deepseek-reasonerdeepseek-reasoner0 exact claim connections · 4 source-qualified statements
  68. provider · deepmind.googleGoogle DeepMind0 exact claim connections · 4 source-qualified statements
  69. legal entity · SpaceXAI LLCSpaceXAI LLC0 exact claim connections · 4 source-qualified statements
  70. governance document · 1.0Artificial Intelligence Risk Management Framework (AI RMF 1.0)0 exact claim connections · 6 source-qualified statements
  71. governance document · methodology-since-2024-11Overview and methodology of the AI Incidents and Hazards Monitor0 exact claim connections · 6 source-qualified statements
  72. provider · cohere.comCohere0 exact claim connections · 4 source-qualified statements
  73. organization · DeepSeek operating organizationDeepSeek0 exact claim connections · 4 source-qualified statements
  74. benchmark · stanford-crfm/helm@63754d05db6f874e41a395880fb573890a13e791Holistic Evaluation of Language Models (HELM)0 exact claim connections · 11 source-qualified statements