AA MMMU-Pro — leaderboard
Artificial Analysis independent evaluation of MMMU-Pro: multimodal understanding across 30 academic disciplines, eliminating shortcut strategies.
Metric: Accuracy (%). Source: artificialanalysis.ai. Status: saturated. 228 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Human Expert | 85.4 |
| 2 | Gemini 3.5 Flash (High) | 84.28 |
| 3 | Gemini 3.5 Flash (Medium) | 83.87 |
| 4 | GPT-5.6 Sol (Max) | 83.41 |
| 5 | GPT-5.6 Sol (xHigh) | 82.66 |
| 6 | Gemini 3.1 Pro (Preview) | 82.43 |
| 7 | GPT-5.6 Sol (High) | 81.85 |
| 8 | GPT-5.6 Sol (Medium) | 81.39 |
| 9 | GPT-5.5 (Medium) | 81.16 |
| 10 | GPT-5.5 (High) | 81.1 |
| 11 | GPT-5.6 Sol (Low) | 80.98 |
| 12 | GPT-5.6 Terra (Max) | 80.69 |
| 13 | Kimi K3 | 80.52 |
| 14 | Muse Spark | 80.52 |
| 15 | Qwen 3.7 Plus | 80.46 |
Interactive version: theaggregate.ai/benchmark?slug=aa-mmmu-pro · How the rankings work · Data refreshed daily, snapshot 2026-07-22.