Calibration
Video-MME (with subtitle) → AA-MMMU-Pro
AA-MMMU-Pro is estimated from Video-MME (with subtitle) with a Michaelis–Menten curve fitted on 8 models measured on both: y = 2.0000·x / (1.38576 + x), R² = 0.50, cross-validated error 2.4 pp. It is used for 0 estimates.
Fitted on
| Model | Video-MME (with subtitle) | AA-MMMU-Pro |
|---|---|---|
| Claude Opus 4.5 | 77.6% | 71.2% |
| Gemini 3 Pro | 88.4% | 80.2% |
| Kimi K2.5 | 87.4% | 75.4% |
| MiniMax M3 | 85.4% | 78.6% |
| Qwen3.5 397B | 87.5% | 77.3% |
| Qwen3.6-27B | 87.7% | 74.6% |
| Qwen3.6-35B-A3B | 86.6% | 75.0% |
| Qwen3.7 Plus | 88.0% | 80.5% |