Model Evaluation and Threat Research (METR) (MEE-tər), is anonprofitresearch institute, based inBerkeley, California,^[1] that evaluatesfrontierAI models' capabilities to carry out long-horizon,agentic tasks that some researchers argue could posecatastrophic risks to society.^[2][3] They have worked with leading AI companies to conduct pre-deployment model evaluations and contribute to system cards, includingOpenAI'so3,o4-mini,GPT-4o andGPT-4.5, andAnthropic'sClaude models.^{[3][4][5][6][7]}

METR'sCEO and founder is Beth Barnes, a formeralignment researcher atOpenAI who left in 2022 to form ARC Evals, the evaluation division ofPaul Christiano'sAlignment Research Center. In December 2023, ARC Evals was thenspun off into an independent501(c)(3) nonprofit and renamed METR.^[8][9][10]

A substantial amount of METR's research is focused on evaluating the capabilities of AI systems to conductresearch and development of AI systems themselves, including RE-Bench, abenchmark designed to test whether AIs can "solve research engineering tasks and accelerate AI R&D".^[11][12]

In March 2025, METR published a paper noting that the length of software engineering tasks that the leading AI model could complete had adoubling time of around 7 months between 2019 and 2024.^[14]

In January 2026, METR has released a new version of their time horizon estimates model (Time Horizon 1.1). According to their new model the rate of progress of AI capabilities has increased since 2023. They now estimate that the post-2023 doubling-time is 130.8 days (4.3 months). Progress is thus estimated to be 20% more rapid.^[15]

METR releases a "task-completion time horizon" for analysed AI models. This measures the "task duration (measured by human expert completion time) at which an AI agent is predicted to succeed with a given level of reliability."^[16] They release it in two variants: The 50%-time horizon, which gives the task duration at which an AI model is estimated to succeed 50% of the time and the 80%-time horizon, which gives the task duration at which an AI model is estimated to succeed 80% of the time.^[16] They have two versions of horizon estimates: Time Horizon 1.1, introduced in January 2026, and the original Time Horizon 1.0.^[16]

As of February 2026^[update] the best performing model isGPT-5.2 (high) with a 6 hours 34 minutes 50%-time horizon and a 80%-time horizon of 55 minutes.^[16] The following table provides the time horizon estimates ordered by the model's release date:^[16]

• Official website

• Non-profit organizations based in California
• 501(c)(3) organizations
• Research institutes in California
• Computer science institutes in the United States
• 2022 establishments in California
• Research institutes established in 2022
• Existential risk from artificial intelligence
• AI safety

• Webarchive template wayback links
• Articles with short description
• Short description matches Wikidata
• Use dmy dates from February 2026
• Pages using infobox mapframe with missing coordinates
• Articles containing potentially dated statements from February 2026
• All articles containing potentially dated statements
• Official website different in Wikidata and Wikipedia