MMMU-Pro: leaderboard
MMMU-Pro evaluates expert-level multimodal understanding with vision and standard variants derived from the MMMU benchmark.
Metric: Accuracy (%). Source: artificialanalysis.ai. 286 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | GPT-6 Astra (Max) | 86.88 |
| 2 | GPT-6 Astra (High) | 86.42 |
| 3 | GPT-6 Astra (xHigh) | 86.24 |
| 4 | GPT-6.1 Sol (Max) | 85.95 |
| 5 | Gemini 3.8 Flash (High) | 85.61 |
| 6 | Gemini 3.7 Flash (High) | 85.49 |
| 7 | GPT-6 Astra (Medium) | 85.09 |
| 8 | GPT-6.1 Sol (xHigh) | 85.09 |
| 9 | GPT-6.1 Sol (High) | 84.91 |
| 10 | Gemini 3.7 Flash (Low) | 84.86 |
| 11 | Claude Opus 5 (Max) | 84.74 |
| 12 | Gemini 3.7 Flash (Medium) | 84.74 |
| 13 | GPT-6 Astra (Low) | 84.62 |
| 14 | Gemini 3.8 Flash (Low) | 84.51 |
| 15 | Gemini 3.5 Flash (High) | 84.28 |
Interactive version: theaggregate.ai/benchmark?slug=mmmu-pro-9jqd0n · How It Works · Data refreshed daily, snapshot 2026-10-09.