HumanEval.orghumans judging AI, in the open

Enter the Arena

Head-to-head matrix

Win-rate grids from eligible votes — pick a category.

Human-Like Chat →Stealth Text →Browser Use →

HumanEval.org

Humans judging AI, in the open. Ratings are computed from human votes with a published, reproducible methodology. Confidence intervals are always shown.

Powered by AI Expert — sponsorship & funding disclosure.

Newsletter

Launch notes and methodology changes — no spam, unsubscribe any time.

Leaderboards

All leaderboards Benchmarks Head-to-head matrix Charts

Models

All models Compare Suggest a model

Participate

Arena Judging queue Become vetted

Research

Articles Analysis Changelog

Data

Downloads Public API

About

Mission Methodology Funding

More

Account RSS feed