OpenEvals - MMLU-Pro: leaderboard
Hugging Face OpenEvals official leaderboard covering MMLU-Pro multi-domain academic reasoning.
Metric: Accuracy (%). Source: huggingface.co. Saturation forecast: Estimated already saturated. 33 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | MiniMax-M2.1 | 88 | #289 |
| 2 | Qwen 3.5 397B A17B | 87.8 | #146 |
| 3 | Kimi K2.5 | 87.1 | #144 |
| 4 | Qwen 3.5 122B A10B | 86.7 | #173 |
| 5 | Qwen 3.5 27B | 86.1 | #187 |
| 6 | Qwen 3.5 35B A3B | 85.3 | #256 |
| 7 | DeepSeek V3.2 | 85 | #205 |
| 8 | DeepSeek R1 0528 | 85 | #222 |
| 9 | Kimi K2 (Thinking) | 84.6 | #243 (Kimi K2) |
| 10 | Qwen 3 235B A22B 2507 (Thinking) | 84.4 | #260 (Qwen 3 235B A22B 2507) |
| 11 | Step 3.5 Flash | 84.4 | #270 |
| 12 | GLM-4.7 | 84.3 | #192 |
| 13 | DeepSeek R1 | 84 | #251 |
| 14 | K-EXAONE | 83.8 | #378 |
| 15 | Qwen 3.5 9B | 82.5 | #379 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=openevals-mmlu-pro · How It Works · Data refreshed daily, snapshot 2026-10-20.