OpenEvals - GPQA Diamond: leaderboard
Hugging Face OpenEvals official leaderboard covering GPQA Diamond graduate-level science reasoning.
Metric: Accuracy (%). Source: huggingface.co. Saturation forecast: Estimated already saturated. 27 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Qwen 3.5 397B A17B | 88.4 | #146 |
| 2 | Kimi K2.5 | 87.6 | #144 |
| 3 | Qwen 3.5 122B A10B | 86.6 | #173 |
| 4 | GLM-5 | 86 | #141 |
| 5 | GLM-4.7 | 85.7 | #192 |
| 6 | Qwen 3.5 27B | 85.5 | #187 |
| 7 | MiniMax-M2.5 | 85.2 | #303 |
| 8 | Kimi K2 (Thinking) | 84.5 | #243 (Kimi K2) |
| 9 | Qwen 3.5 35B A3B | 84.2 | #256 |
| 10 | Nanbeige4.1-3B | 83.8 | #940 |
| 11 | Step 3.5 Flash | 83.5 | #270 |
| 12 | DeepSeek V3.2 | 82.4 | #205 |
| 13 | Qwen 3.5 9B | 81.7 | #379 |
| 14 | LongCat-Flash-Thinking-2601 | 80.5 | #229 |
| 15 | K-EXAONE | 79.1 | #378 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=openevals-gpqa-diamond · How It Works · Data refreshed daily, snapshot 2026-10-20.