Calibration
MMMU-Pro → MMMU-Pro w/ Python
MMMU-Pro w/ Python is estimated from MMMU-Pro with a Hill curve fitted on 9 models measured on both: y = 0.6079 + (1.2000 − 0.6079)·x^6.00 / (0.88881^6.00 + x^6.00), R² = 0.99, cross-validated error 0.6 pp. It is used for 33 estimates.
| Estimated model | MMMU-Pro | MMMU-Pro w/ Python | Source |
|---|---|---|---|
| Claude Opus 4.5 | 70.6% | 72.7% | estimated ± 0.6 pp, high confidence |
| Claude Opus 4.6 | 77.3% | 78.7% | estimated ± 0.6 pp, high confidence |
| Command A+ | 63.0% | 67.5% | estimated ± 0.6 pp, medium confidence |
| dots3-note Preview | 79.1% | 80.4% | estimated ± 0.6 pp, high confidence |
| Gemini 3.1 Pro | 83.9% | 85.3% | estimated ± 0.6 pp, medium confidence |
| Gemini 3.5 Flash | 83.6% | 85.0% | estimated ± 0.6 pp, medium confidence |
| Gemini 3 Pro | 81.0% | 82.4% | estimated ± 0.6 pp, high confidence |
| Gemma 4 12B | 69.1% | 71.5% | estimated ± 0.6 pp, high confidence |
| Gemma 4 26B A4B | 73.8% | 75.4% | estimated ± 0.6 pp, high confidence |
| Gemma 4 31B | 76.9% | 78.3% | estimated ± 0.6 pp, high confidence |
| GPT-5.2 | 79.5% | 80.8% | estimated ± 0.6 pp, high confidence |
| Grok 4.20 | 75.2% | 76.7% | estimated ± 0.6 pp, high confidence |
| Grok 4.3 | 78.1% | 79.5% | estimated ± 0.6 pp, high confidence |
| Inkling | 73.5% | 75.1% | estimated ± 0.6 pp, high confidence |
| Inkling-Small | 74.0% | 75.6% | estimated ± 0.6 pp, high confidence |
| Interfaze Beta | 71.1% | 73.1% | estimated ± 0.6 pp, high confidence |
| Kimi K2.5 | 78.5% | 79.8% | estimated ± 0.6 pp, high confidence |
| Kimi K2.5 (Reasoning) | 78.5% | 79.8% | estimated ± 0.6 pp, high confidence |
| LFM2.5-VL-3B | 30.5% | 60.9% | estimated ± 0.6 pp, medium confidence |
| MiMo-V2.5 | 77.9% | 79.3% | estimated ± 0.6 pp, high confidence |
| MiniMax M3 | 78.1% | 79.5% | estimated ± 0.6 pp, high confidence |
| Muse Glimmer 30B | 74.0% | 75.6% | estimated ± 0.6 pp, high confidence |
| Muse Spark | 80.4% | 81.7% | estimated ± 0.6 pp, high confidence |
| Pareto 26.9 | 78.0% | 79.4% | estimated ± 0.6 pp, high confidence |
| Qwen3.5 397B | 79.0% | 80.3% | estimated ± 0.6 pp, high confidence |
| Qwen3.6-27B | 75.8% | 77.2% | estimated ± 0.6 pp, high confidence |
| Qwen3.6-35B-A3B | 75.3% | 76.8% | estimated ± 0.6 pp, high confidence |
| Qwen3.6 Plus | 78.8% | 80.1% | estimated ± 0.6 pp, high confidence |
| Qwen3.7 Plus | 79.0% | 80.3% | estimated ± 0.6 pp, high confidence |
| Qwen3.8 Max | 82.3% | 83.7% | estimated ± 0.6 pp, high confidence |
| Seed 2.1 Pro | 81.6% | 83.0% | estimated ± 0.6 pp, high confidence |
| Seed 2.1 Turbo | 80.1% | 81.4% | estimated ± 0.6 pp, high confidence |
| Step 5 Preview | 76.0% | 77.4% | estimated ± 0.6 pp, high confidence |