Epoch Capabilities Index
A composite capability index from Epoch AI that statistically stitches together scores from 40+ benchmarks (using an Item-Response-Theory-style model) into a single saturation-resistant general-capability scale, calibrated so Claude 3.5 Sonnet=130 and GPT-5=150.
What this benchmark measures
A composite capability index from Epoch AI that statistically stitches together scores from 40+ benchmarks (using an Item-Response-Theory-style model) into a single saturation-resistant general-capability scale, calibrated so Claude 3.5 Sonnet=130 and GPT-5=150.
Rows on this page are sourced from public benchmark artifacts, leaderboard exports, or source-linked model reports. Each row keeps benchmark version, source model name, and available run details attached to the score.
The metric shown here is Index. It should be interpreted within Epoch Capabilities Index, not compared as part of a site-wide ranking.
Frequently asked
What is Epoch Capabilities Index?
A composite capability index from Epoch AI that statistically stitches together scores from 40+ benchmarks (using an Item-Response-Theory-style model) into a single saturation-resistant general-capability scale, calibrated so Claude 3.5 Sonnet=130 and GPT-5=150. It is a reasoning benchmark measured by Index.
What does Index mean on Epoch Capabilities Index?
Epoch Capabilities Index reports Index; higher is better. Scores are shown only within Epoch Capabilities Index and are never averaged with other benchmarks.
What is the top reported Epoch Capabilities Index score?
GPT-5.5 Pro has the top reported score on Epoch Capabilities Index: 159.3 (Index).
Why do Epoch Capabilities Index scores differ across runs?
Harness, scaffold, reasoning effort, and prompt setup change results, so two runs of the same model can differ. evals.report keeps each score with its run context so the differences stay visible.
Does evals.report rank models across benchmarks?
No. Epoch Capabilities Index scores are shown within their own metric; evals.report never combines benchmarks into a composite ranking or a single "best model".