Humanity's Last Exam: leaderboard
PhD-level questions at the frontier of human expert knowledge. Made by the same team behind MASK. Even top models show high calibration errors, indicating reasoning gaps.
Metric: Score. Source: scale.com. 55 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | GPT-6 Astra | 54.8 | #6 |
| 2 | Claude Fable 5.1 (xHigh) | 46.5 | #9 (Claude Fable 5.1) |
| 3 | Gemini 3.1 Pro (Preview) (High) | 46.44 | #58 (Gemini 3.1 Pro (Preview)) |
| 4 | Claude Sonnet 5.5 (xHigh) | 46.12 | #11 (Claude Sonnet 5.5) |
| 5 | Gemini 3.8 Flash | 44.52 | #21 |
| 6 | Muse Spark | 40.56 | #64 |
| 7 | Gemini 3 Pro (Preview) | 37.52 | #66 |
| 8 | Claude Opus 4.7 | 36.2 | #49 |
| 9 | Claude Opus 4.6 (Adaptive Reasoning, Max Effort) | 34.44 | #63 (Claude Opus 4.6) |
| 10 | Claude Haiku 5.5 | 30.12 | #40 |
| 11 | GPT-5.2 | 27.8 | #110 |
| 12 | GPT-5 | 25.32 | #98 |
| 13 | Claude Opus 4.5 (20251101) (Thinking) | 25.2 | #74 (Claude Opus 4.5 (20251101)) |
| 14 | Kimi K2.5 | 24.37 | #144 |
| 15 | GPT-5.1 (Thinking) | 23.68 | #135 (GPT-5.1) |
No result here: #3 Claude Opus 5.5, #7 GPT-6.1 Sol, #8 Gemini 4 Argon.
Interactive version: theaggregate.ai/benchmark?slug=humanity-s-last-exam · How It Works · Data refreshed daily, snapshot 2026-10-20.