← Humanity's Last Exam leaderboard

Humanity's Last Exam: dated leaderboard snapshots

Permanent, dated copies of the Humanity's Last Exam leaderboard: which model held which rank, at what score, on which version of the test, on what date. One page per result set, each with a CSV.

Scale shows no results date, so each result set is dated by the day ChatJudge first checked it. New snapshots are added when the published results change.

Humanity's Last Exam snapshots
Result set#1Data
First checked 22 Oct 2026GPT 6 Astra 54.80%CSV

Humanity's Last Exam is by the Center for AI Safety (CAIS) and Scale AI; the leaderboard is run and published by Scale AI. ChatJudge copies the published numbers and does not run the benchmark.