AA GPQA Diamond — leaderboard
Artificial Analysis independent evaluation of GPQA Diamond: 198 hardest graduate-level science questions where PhD experts achieve only 65%.
Metric: Accuracy (%). Source: artificialanalysis.ai. Status: saturated. 544 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Gemini 3.1 Pro (Preview) | 94.14 |
| 2 | GPT-5.6 Sol (Max) | 94.14 |
| 3 | GPT-5.5 (xHigh) | 93.54 |
| 4 | Kimi K3 | 93.54 |
| 5 | GPT-5.5 (High) | 93.23 |
| 6 | GPT-5.6 Sol (xHigh) | 93.13 |
| 7 | Grok 4.5 (High) | 93.13 |
| 8 | MiniMax-M3 | 92.93 |
| 9 | GPT-5.6 Sol (High) | 92.83 |
| 10 | GPT-5.5 (Medium) | 92.63 |
| 11 | GPT-5.6 Sol (Medium) | 92.63 |
| 12 | GPT-5.6 Terra (Max) | 92.53 |
| 13 | Qwen 3.7 Max | 92.32 |
| 14 | Gemini 3.5 Flash (High) | 92.22 |
| 15 | Gemini 3.5 Flash (Medium) | 92.12 |
Interactive version: theaggregate.ai/benchmark?slug=aa-gpqa-diamond · How the rankings work · Data refreshed daily, snapshot 2026-07-22.