Humanity's Last Exam
A broad expert-level academic question-answering benchmark for frontier reasoning systems.
What this benchmark measures
A broad expert-level academic question-answering benchmark for frontier reasoning systems.
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 accuracy. It should be interpreted within Humanity's Last Exam, not compared as part of a site-wide ranking.
What to be careful about
Avoid stale scraped tables without retrieved-at metadata.
Frequently asked
What is Humanity's Last Exam?
A broad expert-level academic question-answering benchmark for frontier reasoning systems. It is a reasoning benchmark measured by accuracy.
What does accuracy mean on Humanity's Last Exam?
Humanity's Last Exam reports accuracy (%); higher is better. Scores are shown only within Humanity's Last Exam and are never averaged with other benchmarks.
What is the top reported Humanity's Last Exam score?
Claude Mythos Preview has the top reported score on Humanity's Last Exam: 56.8% (accuracy).
Why do Humanity's Last Exam 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. Humanity's Last Exam scores are shown within their own metric; evals.report never combines benchmarks into a composite ranking or a single "best model".