Calibration
MMMU-Pro → MLVU (M-Avg)
MLVU (M-Avg) is estimated from MMMU-Pro with a linear curve fitted on 10 models measured on both: y = 0.4510·x + 0.5075, R² = 0.85, cross-validated error 2.2 pp. It is used for 22 estimates.
| Estimated model | MMMU-Pro | MLVU (M-Avg) | Source |
|---|---|---|---|
| GPT-6.1 Terra | 85.5% | 89.3% | estimated ± 2.2 pp, medium confidence |
| GPT-5.6 Sol | 83.0% | 88.2% | estimated ± 2.2 pp, medium confidence |
| GPT-5.5 | 81.2% | 87.4% | estimated ± 2.2 pp, high confidence |
| GPT-5.6 Terra | 80.7% | 87.1% | estimated ± 2.2 pp, high confidence |
| Hy4 | 78.6% | 86.2% | estimated ± 2.2 pp, high confidence |
| GPT-5.6 Luna | 78.4% | 86.1% | estimated ± 2.2 pp, high confidence |
| Grok 4.3 | 78.1% | 86.0% | estimated ± 2.2 pp, high confidence |
| MiniMax M3 | 78.1% | 86.0% | estimated ± 2.2 pp, high confidence |
| Pareto 26.9 | 78.0% | 85.9% | estimated ± 2.2 pp, high confidence |
| Claude Opus 4.6 | 77.3% | 85.6% | estimated ± 2.2 pp, high confidence |
| Gemma 4 31B | 76.9% | 85.4% | estimated ± 2.2 pp, high confidence |
| GPT-5.4 mini | 76.6% | 85.3% | estimated ± 2.2 pp, high confidence |
| Step 5 Preview | 76.0% | 85.0% | estimated ± 2.2 pp, high confidence |
| Hy4 flash | 75.8% | 84.9% | estimated ± 2.2 pp, high confidence |
| Gemma 4 26B A4B | 73.8% | 84.0% | estimated ± 2.2 pp, high confidence |
| Interfaze Beta | 71.1% | 82.8% | estimated ± 2.2 pp, high confidence |
| GPT-5.4 nano | 66.1% | 80.6% | estimated ± 2.2 pp, high confidence |
| Gemma 4 E4B | 52.6% | 74.5% | estimated ± 2.2 pp, high confidence |
| Qwen2.5-VL-32B | 49.5% | 73.1% | estimated ± 2.2 pp, high confidence |
| Mistral Small 3.1 | 49.3% | 73.0% | estimated ± 2.2 pp, high confidence |
| Gemma 4 E2B | 44.2% | 70.7% | estimated ± 2.2 pp, medium confidence |
| LFM2.5-VL-3B | 30.5% | 64.5% | estimated ± 2.2 pp, medium confidence |
Fitted on
| Model | MMMU-Pro | MLVU (M-Avg) |
|---|---|---|
| Claude Opus 4.5 | 70.6% | 81.7% |
| Gemini 3 Pro | 81.0% | 83.0% |
| GPT-5.2 | 79.5% | 85.6% |
| Kimi K2.5 | 78.5% | 85.0% |
| LFM3-VL-2B | 45.7% | 71.3% |
| Qwen3.5 397B | 79.0% | 86.7% |
| Qwen3.6-27B | 75.8% | 86.6% |
| Qwen3.6-35B-A3B | 75.3% | 86.2% |
| Qwen3.7 Plus | 79.0% | 87.4% |
| Qwen3.8 Max | 82.3% | 90.8% |