Calibration
SimpleVQA → Video-MME (w/o subtitle)
Video-MME (w/o subtitle) is estimated from SimpleVQA with a Michaelis–Menten curve fitted on 7 models measured on both: y = 1.2778·x / (0.34323 + x), R² = 0.82, cross-validated error 8.1 pp. It is used for 0 estimates.
Fitted on
| Model | SimpleVQA | Video-MME (w/o subtitle) |
|---|---|---|
| Claude Opus 4.5 | 65.7% | 81.4% |
| Gemini 3 Pro | 73.2% | 87.7% |
| GPT-5.2 | 55.8% | 85.8% |
| Kimi K2.5 | 71.2% | 83.2% |
| LFM3-VL-2B | 36.3% | 62.8% |
| Qwen3.5 397B | 67.1% | 83.7% |
| Qwen3.6-35B-A3B | 58.9% | 82.5% |