Calibration
AA-Omniscience Accuracy → MMLU-Pro
MMLU-Pro is estimated from AA-Omniscience Accuracy with a Hill curve fitted on 30 models measured on both: y = 0.6806 + (0.8554 − 0.6806)·x^6.00 / (0.14042^6.00 + x^6.00), R² = 0.56, cross-validated error 5.1 pp. It is used for 63 estimates.
| Estimated model | AA-Omniscience Accuracy | MMLU-Pro | Source |
|---|---|---|---|
| GPT-6.2 Astra | 68.6% | 85.5% | estimated ± 5.1 pp, low confidence |
| GPT-6.1 Sol | 61.8% | 85.5% | estimated ± 5.1 pp, low confidence |
| GPT-6.2 Sol | 61.6% | 85.5% | estimated ± 5.1 pp, low confidence |
| GPT-6 Astra | 61.6% | 85.5% | estimated ± 5.1 pp, low confidence |
| Gemini 4 Carbon | 58.3% | 85.5% | estimated ± 5.1 pp, low confidence |
| GPT-6 Sol | 54.5% | 85.5% | estimated ± 5.1 pp, low confidence |
| Gemini 4 Hydrogen | 50.5% | 85.5% | estimated ± 5.1 pp, low confidence |
| Gemini 4 Argon | 48.3% | 85.5% | estimated ± 5.1 pp, medium confidence |
| GPT-6.1 Terra | 47.6% | 85.5% | estimated ± 5.1 pp, medium confidence |
| DeepSeek V4.1 Flash 1017 | 46.4% | 85.5% | estimated ± 5.1 pp, medium confidence |
| Muse Spark 1.3 | 43.6% | 85.5% | estimated ± 5.1 pp, medium confidence |
| Grok 4.7 | 42.7% | 85.5% | estimated ± 5.1 pp, medium confidence |
| Muse Spark Open | 42.0% | 85.5% | estimated ± 5.1 pp, medium confidence |
| GPT-5.2-Codex | 41.1% | 85.5% | estimated ± 5.1 pp, medium confidence |
| Muse Wonder Full | 41.0% | 85.5% | estimated ± 5.1 pp, medium confidence |
| GPT-5.1-Codex | 39.9% | 85.5% | estimated ± 5.1 pp, medium confidence |
| GPT-5.1-Codex-Max | 39.9% | 85.5% | estimated ± 5.1 pp, medium confidence |
| GPT-6 Luna | 39.6% | 85.5% | estimated ± 5.1 pp, medium confidence |
| Gemini 2.5 Pro | 39.1% | 85.5% | estimated ± 5.1 pp, medium confidence |
| Muse Wonder | 38.0% | 85.5% | estimated ± 5.1 pp, medium confidence |
| Ling 3.0 Tiny | 37.9% | 85.5% | estimated ± 5.1 pp, medium confidence |
| Qwen 3.6 Max (preview) | 37.9% | 85.5% | estimated ± 5.1 pp, medium confidence |
| Step 5 Preview | 36.4% | 85.5% | estimated ± 5.1 pp, medium confidence |
| MiMo-V2.6-Pro | 34.8% | 85.5% | estimated ± 5.1 pp, medium confidence |
| Hy3 Preview | 31.5% | 85.4% | estimated ± 5.1 pp, medium confidence |
| Qwen3.8 Max Preview | 29.5% | 85.3% | estimated ± 5.1 pp, medium confidence |
| GLM-5V-Turbo | 29.3% | 85.3% | estimated ± 5.1 pp, medium confidence |
| GLM-5-Turbo | 28.4% | 85.3% | estimated ± 5.1 pp, medium confidence |
| Apodex 1.1 | 28.2% | 85.3% | estimated ± 5.1 pp, medium confidence |
| Apodex 1.1 Mini | 28.2% | 85.3% | estimated ± 5.1 pp, medium confidence |
| Muse Glimmer 30B | 28.0% | 85.3% | estimated ± 5.1 pp, medium confidence |
| MiMo-V2.6-Flash | 27.2% | 85.2% | estimated ± 5.1 pp, medium confidence |
| MiMo-V2-Pro | 26.6% | 85.2% | estimated ± 5.1 pp, medium confidence |
| Gemini 2.5 Flash | 26.1% | 85.1% | estimated ± 5.1 pp, medium confidence |
| Mistral Large 4 | 26.0% | 85.1% | estimated ± 5.1 pp, medium confidence |
| Step 3.7 Flash | 25.8% | 85.1% | estimated ± 5.1 pp, medium confidence |
| DeepSeek V3 0324 | 24.3% | 84.9% | estimated ± 5.1 pp, medium confidence |
| DeepSeek V3.2 | 24.0% | 84.9% | estimated ± 5.1 pp, medium confidence |
| Grok Code Fast 1 | 23.5% | 84.8% | estimated ± 5.1 pp, medium confidence |
| Claude 4 Sonnet | 22.7% | 84.6% | estimated ± 5.1 pp, medium confidence |
| Mercury 2.5 | 22.0% | 84.4% | estimated ± 5.1 pp, medium confidence |
| Mistral Small 4 | 21.4% | 84.2% | estimated ± 5.1 pp, medium confidence |
| Muse Ember | 21.1% | 84.1% | estimated ± 5.1 pp, medium confidence |
| Magistral Medium | 20.3% | 83.8% | estimated ± 5.1 pp, medium confidence |
| Mistral Large 2 | 19.9% | 83.6% | estimated ± 5.1 pp, medium confidence |
| MiMo-V2-Omni | 19.3% | 83.3% | estimated ± 5.1 pp, medium confidence |
| Solar Pro 3 | 18.5% | 82.7% | estimated ± 5.1 pp, medium confidence |
| North Mini Code | 18.4% | 82.7% | estimated ± 5.1 pp, medium confidence |
| Mistral Medium 3 | 18.3% | 82.6% | estimated ± 5.1 pp, medium confidence |
| Magistral Small | 18.0% | 82.3% | estimated ± 5.1 pp, medium confidence |
| Claude 3 Haiku | 17.6% | 82.0% | estimated ± 5.1 pp, medium confidence |
| A.X K2 | 17.5% | 81.9% | estimated ± 5.1 pp, medium confidence |
| Grok 4.1 Fast | 17.2% | 81.5% | estimated ± 5.1 pp, medium confidence |
| Nova Pro | 16.9% | 81.2% | estimated ± 5.1 pp, medium confidence |
| K-Exaone | 16.4% | 80.6% | estimated ± 5.1 pp, medium confidence |
| GLM-4.5-Air | 16.3% | 80.5% | estimated ± 5.1 pp, medium confidence |
| Solar Pro 2 | 16.1% | 80.2% | estimated ± 5.1 pp, medium confidence |
| Ling 2.6 Flash | 15.6% | 79.5% | estimated ± 5.1 pp, medium confidence |
| Devstral Small | 14.7% | 78.0% | estimated ± 5.1 pp, medium confidence |
| Gemma 3 27B | 13.0% | 74.8% | estimated ± 5.1 pp, medium confidence |
| Command A+ | 9.8% | 69.9% | estimated ± 5.1 pp, medium confidence |
| LFM2.5-8B-A1B | 9.4% | 69.5% | estimated ± 5.1 pp, medium confidence |
| LFM3-1.7B | 7.1% | 68.3% | estimated ± 5.1 pp, low confidence |
Fitted on
| Model | AA-Omniscience Accuracy | MMLU-Pro |
|---|---|---|
| Claude 3.5 Haiku | 13.2% | 65.0% |
| Claude Opus 4.5 | 40.9% | 89.5% |
| Claude Opus 4.6 | 45.8% | 82.0% |
| Claude Sonnet 4.6 | 38.6% | 79.2% |
| DeepSeek-R1 | 30.5% | 84.0% |
| DeepSeek V3 | 25.5% | 75.9% |
| DeepSeek V3.1 | 23.1% | 83.7% |
| DeepSeek V4 Flash 0731 | 40.4% | 86.2% |
| DeepSeek V4 Pro | 49.1% | 87.5% |
| Gemma 4 12B | 15.7% | 77.2% |
| Gemma 4 31B | 19.9% | 85.2% |
| Gemma 4 E4B | 9.7% | 69.4% |
| GLM-4.7 | 29.3% | 84.3% |
| GLM-5 | 26.3% | 85.7% |
| Granite 4.2 30B | 11.2% | 77.6% |
| Granite 4.2 3B | 9.2% | 67.8% |
| Granite 4.2 8B | 11.8% | 74.0% |
| K-EXAONE 2.0 | 14.5% | 83.5% |
| Kimi K2 | 27.4% | 81.1% |
| Kimi K2.5 | 35.2% | 87.1% |
| MiMo-V2-Flash | 15.6% | 84.9% |
| Mistral Small 3.1 | 15.1% | 66.8% |
| Nemotron 3.5 Lightning 30B A3B NVFP4 | 14.7% | 81.6% |
| Nemotron 3 Ultra | 21.6% | 86.8% |
| Qwen3.5-27B | 20.7% | 86.1% |
| Qwen3.5-35B-A3B | 20.1% | 85.3% |
| Qwen3.6-27B | 19.6% | 86.2% |
| Qwen3.6 Plus | 26.4% | 88.5% |
| Qwen3.7 Max | 31.1% | 89.6% |
| Solar Pro 4 | 18.4% | 86.3% |