Humanity's Last Exam
2,500 expert-written questions across dozens of fields, without tools.
| Category | Knowledge |
|---|---|
| Metric | Accuracy |
| Source | Epoch AI (evaluator) |
| License | CC BY 4.0 |
| Benchmark page | lastexam.ai |
| Status | Scored |
| Weight | 1.2 within Knowledge |
| Fit | Discrimination 0.67, noise 0.11 |
Results
| # | Model | Provider | Result | Date |
|---|---|---|---|---|
| 1 | GPT-6 Astra | OpenAI | 54.8% | 2026-11-11 |
| 2 | Claude Fable 5.1 Extra High | Anthropic | 46.5% | 2026-11-11 |
| 3 | Gemini 3.1 Pro Preview | 46.4% | 2026-11-11 | |
| 4 | Claude Sonnet 5.5 Extra High | Anthropic | 46.1% | 2026-11-11 |
| 5 | Gemini 3.8 Flash | 44.5% | 2026-11-11 | |
| 6 | GPT-5.4 Pro | OpenAI | 44.3% | 2026-11-11 |
| 7 | Muse Spark | Meta | 40.6% | 2026-11-11 |
| 8 | Gemini 3 Pro Preview | 37.5% | 2026-11-11 | |
| 9 | GPT-5.4 Extra High | OpenAI | 36.2% | 2026-11-11 |
| 10 | Claude Opus 4.7 | Anthropic | 36.2% | 2026-11-11 |
| 11 | Claude Opus 4.6 Max | Anthropic | 34.4% | 2026-11-11 |
| 12 | GPT-5 Pro | OpenAI | 31.6% | 2026-11-11 |
| 13 | Claude Haiku 5.5 | Anthropic | 30.1% | 2026-11-11 |
| 14 | GPT-5.2 | OpenAI | 27.8% | 2026-11-11 |
| 15 | GPT-5 | OpenAI | 25.3% | 2026-11-11 |
| 16 | GPT-5 High | OpenAI | 25.3% | 2026-11-11 |
| 17 | Claude Opus 4.5 | Anthropic | 25.2% | 2026-11-11 |
| 18 | Kimi K2.5 | Moonshot AI | 24.4% | 2026-11-11 |
| 19 | GPT-5.1 | OpenAI | 23.7% | 2026-11-11 |
| 20 | Gemini 2.5 Pro Preview 06-05 | 21.6% | 2026-11-11 | |
| 21 | o3 High | OpenAI | 20.3% | 2026-11-11 |
| 22 | GPT-5 Mini | OpenAI | 19.4% | 2026-11-11 |
| 23 | o3 Medium | OpenAI | 19.2% | 2026-11-11 |
| 24 | Claude Opus 4.6 | Anthropic | 19.0% | 2026-11-11 |
| 25 | o4-mini High | OpenAI | 18.1% | 2026-11-11 |
| 26 | o4-mini Medium | OpenAI | 14.3% | 2026-11-11 |
| 27 | Claude Sonnet 4.5 | Anthropic | 13.7% | 2026-11-11 |
| 28 | Gemini 2.5 Flash (Apr 2025) | 12.1% | 2026-11-11 | |
| 29 | Claude Opus 4.1 | Anthropic | 11.5% | 2026-11-11 |
| 30 | Gemini 2.5 Flash Preview 05-20 | 11.0% | 2026-11-11 | |
| 31 | Claude Opus 4 | Anthropic | 10.7% | 2026-11-11 |
| 32 | Gemini 3.1 Flash Lite | 8.6% | 2026-11-11 | |
| 33 | GLM-4.5 | Zhipu AI | 8.3% | 2026-11-11 |
| 34 | GLM-4.5-Air | Zhipu AI | 8.1% | 2026-11-11 |
| 35 | o1-pro | OpenAI | 8.1% | 2026-11-11 |
| 36 | Claude Sonnet 3.7 | Anthropic | 8.0% | 2026-11-11 |
| 37 | o1 | OpenAI | 8.0% | 2026-11-11 |
| 38 | Claude Sonnet 4 | Anthropic | 7.8% | 2026-11-11 |
| 39 | GPT-5.1 Non-reasoning | OpenAI | 6.8% | 2026-11-11 |
| 40 | Gemini 2.0 Flash Thinking (Jan 2025) | 6.6% | 2026-11-11 | |
| 41 | Llama 4 Maverick | Meta | 5.7% | 2026-11-11 |
| 42 | GPT-4.5 | OpenAI | 5.4% | 2026-11-11 |
| 43 | GPT-4.1 | OpenAI | 5.4% | 2026-11-11 |
| 44 | Gemini 1.5 Pro (Sept 2024) | 4.6% | 2026-11-11 | |
| 45 | Mistral Medium 3 | Mistral | 4.5% | 2026-11-11 |
| 46 | Amazon Nova Pro | Amazon | 4.4% | 2026-11-11 |
| 47 | Claude Sonnet 3.5 v2 | Anthropic | 4.1% | 2026-11-11 |
| 48 | Amazon Nova Lite | Amazon | 3.6% | 2026-11-11 |
| 49 | GPT-4o (2024-11-20) | OpenAI | 2.7% | 2026-11-11 |