Epoch Capabilities Index

A composite index from Epoch AI that combines results across many benchmarks into a single score of an AI model's overall capability.

4 articlesLast mentioned

The Epoch Capabilities Index (ECI) is a composite measure of AI model capability published by the research organization Epoch AI in 2025. It uses a statistical model to place results from many different benchmarks on a single scale and assigns each model a score. The index was designed to compensate for individual benchmarks saturating or covering different ranges of difficulty, so that models from different generations can be compared on one axis.

The ECI is used to analyze trends in AI capability growth over time, and changes in the scores of top models are frequently cited in discussions of the pace of AI progress.

This entry is based on AIPOST articles and widely known facts. If something is wrong, please send us a correction request.

Articles covering this entry

  • GPT-6.2 Astra: Cheaper per Task and Strong in 3D, but Not the Top ModelAI News ·

    On the OpenAI Epoch Capabilities Index, Astra scores 170.1 and Sol 168.8, in line with a trend of roughly 15.6 points per year since GPT-5.

    Read article
  • METR Audit Finds Anthropic AI R&D Running at Twice Its Historical PaceAI News ·

    On Anthropic's internal capability index, called AECI (Anthropic Epoch Capabilities Index), METR reports an acceleration factor of 1.9x since October 2025, with…

    Read article
  • Anthropic S-1 RSI Disclosure: 2x Research Speedup, Not a Runaway LoopAI News ·

    Epoch Capabilities Index equivalent About 10 percent Approximate researcher productivity gain per unit on Epoch AI's capabilities milestone index

    Read article
  • METR Report: AI Progress Doubles in Speed, Raising 2027 Self-Improvement DebateAI News ·

    …rimary public evidence rests on Anthropic's internal version of the Epoch Capabilities Index, known as the AECI, while Epoch's own public ECI shows a relatively…

    Read article

Found a mistake or missing an entry?

Wiki entries are based on AIPOST articles and widely known facts. If an entry is wrong, send us a correction request and we will check and fix it. Names from new articles are added to the wiki after the article is published.