METR - Wikipedia

reference

Wikipedia·en.wikipedia.org/wiki/METR

Credibility Rating

3/5

Good(3)

Good quality. Reputable source with community review or editorial standards, but less rigorous than peer-reviewed venues.

Rating inherited from publication venue: Wikipedia

Data Status

Not fetched

Cited by 1 page

Page	Type	Quality
METR	Organization	66.0

Cached Content Preview

HTTP 200Fetched Feb 27, 202629 KB

[Jump to content](https://en.wikipedia.org/wiki/METR#bodyContent)

From Wikipedia, the free encyclopedia

AI model evaluation nonprofit

| Formation | 2022; 4 years ago (2022) |
| Founder | Beth Barnes |
| Type | [Nonprofit](https://en.wikipedia.org/wiki/Nonprofit "Nonprofit") [research institute](https://en.wikipedia.org/wiki/Research_institute "Research institute") |
| Legal status | [501(c)(3)](https://en.wikipedia.org/wiki/501(c)(3) "501(c)(3)") [tax exempt](https://en.wikipedia.org/wiki/Tax_exempt "Tax exempt") [charity](https://en.wikipedia.org/wiki/Charitable_organization "Charitable organization") |
| Purpose | AI [safety research](https://en.wikipedia.org/wiki/AI_safety "AI safety") and model evaluation |
| Location | - [Berkeley, California](https://en.wikipedia.org/wiki/Berkeley,_California "Berkeley, California") |
| Website | [metr.org](https://metr.org/) |

METR

**Model Evaluation and Threat Research** ( **METR**) ( [_MEE-tər_](https://en.wikipedia.org/wiki/Help:Pronunciation_respelling_key "Help:Pronunciation respelling key")), is a [nonprofit](https://en.wikipedia.org/wiki/Nonprofit_organization "Nonprofit organization") [research institute](https://en.wikipedia.org/wiki/Research_institute "Research institute"), based in [Berkeley, California](https://en.wikipedia.org/wiki/Berkeley,_California "Berkeley, California"),[\[1\]](https://en.wikipedia.org/wiki/METR#cite_note-1) that evaluates [frontier](https://en.wikipedia.org/wiki/Frontier_model "Frontier model") [AI](https://en.wikipedia.org/wiki/Artificial_intelligence "Artificial intelligence") models' capabilities to carry out long-horizon, [agentic](https://en.wikipedia.org/wiki/Agentic_AI "Agentic AI") tasks that some researchers argue could pose [catastrophic risks to society](https://en.wikipedia.org/wiki/Existential_risk_from_artificial_intelligence "Existential risk from artificial intelligence").[\[2\]](https://en.wikipedia.org/wiki/METR#cite_note-2)[\[3\]](https://en.wikipedia.org/wiki/METR#cite_note-:0-3) They have worked with leading AI companies to conduct pre-deployment model evaluations and contribute to system cards, including [OpenAI](https://en.wikipedia.org/wiki/OpenAI "OpenAI")'s [o3](https://en.wikipedia.org/wiki/OpenAI_o3 "OpenAI o3"), [o4-mini](https://en.wikipedia.org/wiki/OpenAI_o4-mini "OpenAI o4-mini"), [GPT-4o](https://en.wikipedia.org/wiki/GPT-4o "GPT-4o") and [GPT-4.5](https://en.wikipedia.org/wiki/GPT-4.5 "GPT-4.5"), and [Anthropic](https://en.wikipedia.org/wiki/Anthropic "Anthropic")'s [Claude](https://en.wikipedia.org/wiki/Claude_(language_model) "Claude (language model)") models.[\[3\]](https://en.wikipedia.org/wiki/METR#cite_note-:0-3)[\[4\]](https://en.wikipedia.org/wiki/METR#cite_note-4)[\[5\]](https://en.wikipedia.org/wiki/METR#cite_note-5)[\[6\]](https://en.wikipedia.org/wiki/METR#cite_note-6)[\[7\]](https://en.wikipedia.org/wiki/METR#cite_note-7)

METR's [CEO](https://en.wikipedia.org/wiki/Chief_executive_officer "Chief executive officer") an

... (truncated, 29 KB total)

Resource ID: ab9cc01cf367fd79 | Stable ID: YzYwODZlOD