Skip to content
K = 0.731
Benchmark sources

Humanity’s Last Exam

CAIS / Scale AI / Knowledge · Difficult academic knowledge and reasoning across disciplines.
Official benchmark
Across e/accUnder observation
Evidence budgetNot scored
Upstream data timeUnverified
Last successful syncCollection not started

What it measures and how

Fix tasks, tool conditions and judges. Compare HLE, HLE-Rolling and tool-assisted results separately. AA v4.3 already includes HLE; any usable specialized results would serve categories only, without duplicate overall votes.

How this evidence is used

Under observation: the organization has added some new models, but raw scores still lack verifiable evaluation dates. Epoch’s self-evaluation license is not substituted, and existing AA aggregate evidence is not counted again.

Limitations and attribution

Under observation: the organization has added some new models, but raw scores still lack verifiable evaluation dates. Epoch’s self-evaluation license is not substituted, and existing AA aggregate evidence is not counted again.

Data license: Evaluation code is MIT. Display terms for official leaderboard results remain unconfirmed.

Results are published by CAIS / Scale AI. Raw scores and the e/acc consensus index use different scales and cannot be added directly.