Humanity's Last Exam
knowledgeAround 2,500 expert-written, closed-ended questions across 100+ academic subjects. The hardest broad-knowledge exam in use; frontier models still score well below human experts.
Official benchmark pageModel rankings on Humanity's Last Exam
| # | Model | Score | As of | Source |
|---|---|---|---|---|
| 1 |  Claude Opus 4.8 | 49.8% | May 28, 2026 | cite |
| 2 | Claude Opus 4.7 | 46.9% | May 28, 2026 | cite |
| 3 |  Gemini 3.1 Pro Preview | 44.4% | May 28, 2026 | cite |
| 4 |  GPT-5.5 Pro | 43.1% | May 28, 2026 | cite |
| 5 | GPT-5.5 | 41.4% | May 28, 2026 | cite |
| 6 |  GLM 5.2 | 40.5% | Jun 13, 2026 | cite |
| 7 |  DeepSeek V4 Pro | 37.7% | Apr 24, 2026 | cite |
| 8 | DeepSeek V4 Flash | 34.8% | Apr 24, 2026 | cite |
| 9 | Gemini 3 Flash Preview | 33.7% | Dec 17, 2025 | cite |
| 10 | Claude Sonnet 4.6 | 33.2% | Feb 17, 2026 | cite |
| 11 |  Kimi K2 Thinking | 23.9% | Nov 6, 2025 | cite |
| 12 | Gemini 2.5 Pro | 21.6% | Jun 27, 2025 | cite |
Scores are self-reported or from primary evaluations, each linked to its source. Test conditions (tools, shots, prompt) vary between labs — see the source for details.