Calibration
AA-MMMU-Pro → OfficeQA Pro
OfficeQA Pro is estimated from AA-MMMU-Pro with a offset logistic curve fitted on 9 models measured on both: y = 0.5032 + (0.6762 − 0.5032) / (1 + exp(−200.00·(x − 0.8033))), R² = 0.71, cross-validated error 6.8 pp. It is used for 67 estimates.
| Estimated model | AA-MMMU-Pro | OfficeQA Pro | Source |
|---|---|---|---|
| GPT-6.2 Astra | 89.5% | 67.6% | estimated ± 6.8 pp, low confidence |
| GPT-6 Astra | 86.9% | 67.6% | estimated ± 6.8 pp, medium confidence |
| GPT-6.1 Sol | 86.0% | 67.6% | estimated ± 6.8 pp, medium confidence |
| Gemini 3.8 Flash | 85.6% | 67.6% | estimated ± 6.8 pp, medium confidence |
| Gemini 3.7 Flash | 85.5% | 67.6% | estimated ± 6.8 pp, medium confidence |
| GPT-6.1 Terra | 84.7% | 67.6% | estimated ± 6.8 pp, medium confidence |
| Gemini 3.5 Flash | 84.3% | 67.6% | estimated ± 6.8 pp, medium confidence |
| GPT-5.6 Sol | 83.4% | 67.6% | estimated ± 6.8 pp, medium confidence |
| Gemini 3.6 Flash | 83.2% | 67.6% | estimated ± 6.8 pp, medium confidence |
| GPT-6 Sol | 82.9% | 67.5% | estimated ± 6.8 pp, medium confidence |
| Qwen3.8 Max Preview | 82.8% | 67.5% | estimated ± 6.8 pp, medium confidence |
| Gemini 3.1 Pro | 82.4% | 67.3% | estimated ± 6.8 pp, medium confidence |
| Muse Spark Open | 80.9% | 63.4% | estimated ± 6.8 pp, medium confidence |
| GPT-5.6 Terra | 80.7% | 62.0% | estimated ± 6.8 pp, medium confidence |
| Muse Spark | 80.5% | 60.4% | estimated ± 6.8 pp, medium confidence |
| Qwen3.7 Plus | 80.5% | 60.4% | estimated ± 6.8 pp, medium confidence |
| Grok 4.5 | 80.4% | 59.5% | estimated ± 6.8 pp, medium confidence |
| Gemini 3 Pro | 80.2% | 57.8% | estimated ± 6.8 pp, medium confidence |
| GPT-6 Luna | 79.7% | 54.1% | estimated ± 6.8 pp, medium confidence |
| Kimi K2.6 | 79.4% | 52.6% | estimated ± 6.8 pp, medium confidence |
| Apodex 1.1 | 79.2% | 51.9% | estimated ± 6.8 pp, medium confidence |
| Apodex 1.1 Mini | 79.2% | 51.9% | estimated ± 6.8 pp, medium confidence |
| MiniMax M3.1 Flash | 79.1% | 51.7% | estimated ± 6.8 pp, medium confidence |
| Gemini 3.5 Flash-Lite | 79.0% | 51.4% | estimated ± 6.8 pp, medium confidence |
| Ling 3.0 Flash VL | 79.0% | 51.4% | estimated ± 6.8 pp, medium confidence |
| Gemini 3 Flash | 78.6% | 50.8% | estimated ± 6.8 pp, medium confidence |
| GPT-5.6 Luna | 78.6% | 50.8% | estimated ± 6.8 pp, medium confidence |
| Qwen3.8-Flash-Next | 78.6% | 50.8% | estimated ± 6.8 pp, medium confidence |
| GPT-5.3 Codex | 78.5% | 50.7% | estimated ± 6.8 pp, medium confidence |
| Grok 4.3 | 78.1% | 50.5% | estimated ± 6.8 pp, medium confidence |
| Qwen3.6 Plus | 78.0% | 50.5% | estimated ± 6.8 pp, medium confidence |
| Qwen3.5 397B | 77.3% | 50.4% | estimated ± 6.8 pp, medium confidence |
| GPT-5.2-Codex | 76.3% | 50.3% | estimated ± 6.8 pp, low confidence |
| Qwen3.8-27B | 76.3% | 50.3% | estimated ± 6.8 pp, low confidence |
| GPT-5.1 | 75.5% | 50.3% | estimated ± 6.8 pp, low confidence |
| Claude Opus 4.6 (Adaptive) | 75.4% | 50.3% | estimated ± 6.8 pp, low confidence |
| Kimi K2.5 | 75.4% | 50.3% | estimated ± 6.8 pp, low confidence |
| Step 3.7 Flash | 75.3% | 50.3% | estimated ± 6.8 pp, low confidence |
| Qwen3.5-122B-A10B | 75.0% | 50.3% | estimated ± 6.8 pp, low confidence |
| Qwen3.5-27B | 75.0% | 50.3% | estimated ± 6.8 pp, low confidence |
| Qwen3.6-35B-A3B | 75.0% | 50.3% | estimated ± 6.8 pp, low confidence |
| Gemini 2.5 Pro | 74.9% | 50.3% | estimated ± 6.8 pp, low confidence |
| Qwen3.6-27B | 74.6% | 50.3% | estimated ± 6.8 pp, low confidence |
| Muse Glimmer 30B | 74.3% | 50.3% | estimated ± 6.8 pp, low confidence |
| GPT-5 | 74.2% | 50.3% | estimated ± 6.8 pp, low confidence |
| Claude Opus 4.5 Thinking | 74.0% | 50.3% | estimated ± 6.8 pp, low confidence |
| Inkling-Small | 74.0% | 50.3% | estimated ± 6.8 pp, low confidence |
| Inkling | 73.5% | 50.3% | estimated ± 6.8 pp, low confidence |
| Gemma 4 31B | 73.4% | 50.3% | estimated ± 6.8 pp, low confidence |
| GPT-5.4 mini | 73.3% | 50.3% | estimated ± 6.8 pp, low confidence |
| MiMo-V2.6-Flash | 73.1% | 50.3% | estimated ± 6.8 pp, low confidence |
| GLM-5V-Turbo | 72.8% | 50.3% | estimated ± 6.8 pp, low confidence |
| Qwen3.5-35B-A3B | 72.7% | 50.3% | estimated ± 6.8 pp, low confidence |
| Claude Opus 4.6 | 72.5% | 50.3% | estimated ± 6.8 pp, low confidence |
| GPT-5.1-Codex | 72.5% | 50.3% | estimated ± 6.8 pp, low confidence |
| GPT-5.1-Codex-Max | 72.5% | 50.3% | estimated ± 6.8 pp, low confidence |
| Claude Opus 4.5 | 71.2% | 50.3% | estimated ± 6.8 pp, low confidence |
| Claude Sonnet 4.6 | 70.6% | 50.3% | estimated ± 6.8 pp, low confidence |
| o3 | 70.1% | 50.3% | estimated ± 6.8 pp, low confidence |
| MiMo-V2-Omni | 69.9% | 50.3% | estimated ± 6.8 pp, low confidence |
| Gemma 4 12B | 69.7% | 50.3% | estimated ± 6.8 pp, low confidence |
| Gemma 4 26B A4B | 69.2% | 50.3% | estimated ± 6.8 pp, low confidence |
| o4-mini | 69.2% | 50.3% | estimated ± 6.8 pp, low confidence |
| Grok 4 | 68.8% | 50.3% | estimated ± 6.8 pp, low confidence |
| Gemini 2.5 Flash | 65.5% | 50.3% | estimated ± 6.8 pp, low confidence |
| GPT-5.4 nano | 65.4% | 50.3% | estimated ± 6.8 pp, low confidence |
| Qwen3.5-4B | 65.4% | 50.3% | estimated ± 6.8 pp, low confidence |
Fitted on
| Model | AA-MMMU-Pro | OfficeQA Pro |
|---|---|---|
| Claude Opus 4.7 | 78.8% | 43.6% |
| Claude Opus 5 | 84.7% | 66.9% |
| Claude Opus 5.5 | 87.7% | 67.7% |
| GPT-5.4 | 78.4% | 53.2% |
| GPT-5.5 | 79.9% | 54.1% |
| Kimi K3.1 Pro | 83.9% | 67.4% |
| Kimi K3 | 80.5% | 63.3% |
| MiniMax M3 | 78.6% | 45.1% |
| Step 5 Preview | 76.4% | 60.3% |