Calibration
We-Math → Video-MME (with subtitle)
Video-MME (with subtitle) is estimated from We-Math with a Hill curve fitted on 6 models measured on both: y = 0.0000 + (0.9526 − 0.0000)·x^6.00 / (0.56565^6.00 + x^6.00), R² = 0.91, cross-validated error 4.0 pp. It is used for 0 estimates.
Fitted on
| Model | We-Math | Video-MME (with subtitle) |
|---|---|---|
| Claude Opus 4.5 | 70.0% | 77.6% |
| Gemini 3 Pro | 86.9% | 88.4% |
| GPT-5.2 | 79.0% | 86.0% |
| Kimi K2.5 | 84.7% | 87.4% |
| LFM3-VL-2B | 68.6% | 69.0% |
| Qwen3.5 397B | 87.9% | 87.5% |