Every AI leaderboard

Leaderboards28Models ranked986

28 leaderboards · 986 models ranked

Humanity's Last Exam leaderboard: who is #1

Scale AI and CAIS · Accuracy, higher is better · 44 models · published Nov 9, 2026

Scale AI and CAIS

Top 25 of 44 models, best variant per model as published

#ModelScoreCorrect
1GPT-6 Astraas “GPT 6 Astra”54.8%
2Claude Fable 5.1as “Fable 5.1 (xhigh)”46.5%
3Gemini 3.1 Pro Previewas “gemini-3.1-pro-preview (thinking high)”46.4%
4New, entered at #4 since Oct 13, 2026Claude Sonnet 5.5as “Sonnet 5.5 (xhigh)”46.1%
51down from #4 since Oct 13, 2026Gemini 3.8 FlashGoogle44.5%
61down from #5 since Oct 13, 2026GPT-5.4 Proas “gpt-5.4-pro-2026-03-05”44.3%
71down from #6 since Oct 13, 2026Muse SparkMeta40.6%
81down from #7 since Oct 13, 2026Gemini 3 Proas “gemini-3-pro-preview”37.5%
91down from #8 since Oct 13, 2026GPT-5.4as “gpt-5.4-2026-03-05 (xhigh thinking)”36.2%
101down from #9 since Oct 13, 2026Claude Opus 4.7as “claude-opus-4-7”36.2%
111down from #10 since Oct 13, 2026Claude Opus 4.6as “claude-opus-4-6-thinking-max”34.4%
121down from #11 since Oct 13, 2026GPT-5 Proas “gpt-5-pro-2025-10-06”31.6%
131down from #12 since Oct 13, 2026Claude Haiku 5.5Anthropic30.1%
141down from #13 since Oct 13, 2026GPT-5.2as “gpt-5.2-2025-12-11”27.8%
151down from #14 since Oct 13, 2026GPT-5as “gpt-5-2025-08-07”25.3%
161down from #15 since Oct 13, 2026Claude Opus 4.5as “claude-opus-4-5-20251101-thinking”25.2%
171down from #16 since Oct 13, 2026Kimi K2.5as “kimi-k2.5”24.4%
181down from #17 since Oct 13, 2026GPT-5.1as “gpt-5.1-thinking”23.7%
191down from #18 since Oct 13, 2026Gemini 2.5 Pro (Jun 2025)as “gemini-2.5-pro-preview-06-05”21.6%
201down from #19 since Oct 13, 2026o3as “o3 (high) (April 2025)”20.3%
211down from #20 since Oct 13, 2026GPT-5 minias “gpt-5-mini-2025-08-07”19.4%
221down from #21 since Oct 13, 2026o4-minias “o4-mini (high) (April 2025)”18.1%
231down from #22 since Oct 13, 2026Claude Sonnet 4.5as “claude-sonnet-4-5-20250929-thinking”13.7%
241down from #23 since Oct 13, 2026Gemini 2.5 Flash (Sep 2025)as “Gemini 2.5 Flash (April 2025)”12.1%
251down from #24 since Oct 13, 2026Claude Opus 4.1as “claude-opus-4-1-20250805-thinking”11.5%
About this board and its history

What it measures

2,500 expert-written questions across dozens of subjects, built to be the last closed-ended academic exam.

Rank at each month's end

12345678910SepOctNow1GPT-6 Astra2Claude Fable 5.13Gemini 3.1 Pro Preview4Claude Sonnet 5.55Gemini 3.8 Flash6GPT-5.4 Pro
GPT-6 AstraClaude Fable 5.1Gemini 3.1 Pro PreviewClaude Sonnet 5.5Gemini 3.8 FlashGPT-5.4 Pro

Other reasoning boards

Source

Scores as published by Scale AI and CAIS on the capture date (2026-11-09). Source: labs.scale.com leaderboard page.

labs.scale.com/leaderboard/humanitys_last_exam