Calibration
AA-Omniscience Accuracy → MMLU-Pro
MMLU-Pro is estimated from AA-Omniscience Accuracy with a offset logistic curve fitted on 36 models measured on both: y = 0.0000 + (0.8574 − 0.0000) / (1 + exp(−28.24·(x − 0.0340))), R² = 0.77, cross-validated error 3.5 pp. It is used for 108 estimates.
| Estimated model | AA-Omniscience Accuracy | MMLU-Pro | Source |
|---|---|---|---|
| A.X K2 | 18.6% | 84.6% | estimated ± 3.5 pp, high confidence |
| Apodex 1.1 | 31.7% | 85.7% | estimated ± 3.5 pp, high confidence |
| Apodex 1.1 Mini | 31.7% | 85.7% | estimated ± 3.5 pp, high confidence |
| Claude 3 Haiku | 17.6% | 84.2% | estimated ± 3.5 pp, high confidence |
| Claude 4 Sonnet | 22.7% | 85.4% | estimated ± 3.5 pp, high confidence |
| Claude Fable 5 | 65.4% | 85.7% | estimated ± 3.5 pp, medium confidence |
| Claude Opus 4.5 Thinking | 46.6% | 85.7% | estimated ± 3.5 pp, high confidence |
| Claude Opus 4.6 (Adaptive) | 47.0% | 85.7% | estimated ± 3.5 pp, high confidence |
| Claude Opus 4.7 | 44.7% | 85.7% | estimated ± 3.5 pp, high confidence |
| Command A+ | 8.9% | 70.8% | estimated ± 3.5 pp, high confidence |
| DeepSeek-R1 | 30.5% | 85.7% | estimated ± 3.5 pp, high confidence |
| DeepSeek V3 0324 | 24.3% | 85.5% | estimated ± 3.5 pp, high confidence |
| DeepSeek V3.1 | 23.1% | 85.4% | estimated ± 3.5 pp, high confidence |
| DeepSeek V3.1 (Reasoning) | 29.0% | 85.7% | estimated ± 3.5 pp, high confidence |
| DeepSeek V3.2 | 24.0% | 85.5% | estimated ± 3.5 pp, high confidence |
| Exaone 4.0 1.2B | 5.0% | 52.4% | estimated ± 3.5 pp, medium confidence |
| Gemini 2.5 Flash | 26.1% | 85.6% | estimated ± 3.5 pp, high confidence |
| Gemini 3.5 Flash-Lite | 29.5% | 85.7% | estimated ± 3.5 pp, high confidence |
| Gemini 3.6 Flash | 50.0% | 85.7% | estimated ± 3.5 pp, medium confidence |
| Gemini 3.7 Flash | 55.3% | 85.7% | estimated ± 3.5 pp, medium confidence |
| Gemini 3.8 Flash | 54.6% | 85.7% | estimated ± 3.5 pp, medium confidence |
| Gemini 3 Flash | 45.8% | 85.7% | estimated ± 3.5 pp, high confidence |
| Gemini 3 Pro | 55.8% | 85.7% | estimated ± 3.5 pp, medium confidence |
| Gemini 4 Argon | 49.9% | 85.7% | estimated ± 3.5 pp, medium confidence |
| Gemma 3 27B | 13.0% | 80.4% | estimated ± 3.5 pp, high confidence |
| GLM-4.5-Air | 16.3% | 83.6% | estimated ± 3.5 pp, high confidence |
| GLM-4.6 | 21.4% | 85.2% | estimated ± 3.5 pp, high confidence |
| GLM-5.1 | 23.7% | 85.5% | estimated ± 3.5 pp, high confidence |
| GLM-5.3 | 33.9% | 85.7% | estimated ± 3.5 pp, high confidence |
| GLM-5-Turbo | 28.4% | 85.7% | estimated ± 3.5 pp, high confidence |
| GLM-5V-Turbo | 29.3% | 85.7% | estimated ± 3.5 pp, high confidence |
| GPT-4o | 19.9% | 84.9% | estimated ± 3.5 pp, high confidence |
| GPT-5.1 | 37.7% | 85.7% | estimated ± 3.5 pp, high confidence |
| GPT-5.1-Codex | 39.9% | 85.7% | estimated ± 3.5 pp, high confidence |
| GPT-5.1-Codex-Max | 39.9% | 85.7% | estimated ± 3.5 pp, high confidence |
| GPT-5.2-Codex | 41.1% | 85.7% | estimated ± 3.5 pp, high confidence |
| GPT-5.3 Codex | 52.9% | 85.7% | estimated ± 3.5 pp, medium confidence |
| GPT-5 (high) | 40.3% | 85.7% | estimated ± 3.5 pp, high confidence |
| GPT-5 (medium) | 39.5% | 85.7% | estimated ± 3.5 pp, high confidence |
| GPT-6.1 Sol | 62.1% | 85.7% | estimated ± 3.5 pp, medium confidence |
| GPT-6 Luna | 43.8% | 85.7% | estimated ± 3.5 pp, high confidence |
| GPT-6 Sol | 54.5% | 85.7% | estimated ± 3.5 pp, medium confidence |
| GPT-OSS 120B | 21.8% | 85.3% | estimated ± 3.5 pp, high confidence |
| GPT-OSS 20B | 16.0% | 83.4% | estimated ± 3.5 pp, high confidence |
| Granite-4.0-350M | 3.9% | 45.9% | estimated ± 3.5 pp, medium confidence |
| Granite-4.0-H-1B | 5.2% | 53.5% | estimated ± 3.5 pp, medium confidence |
| Granite-4.0-H-350M | 3.8% | 45.3% | estimated ± 3.5 pp, medium confidence |
| Grok 4 | 40.5% | 85.7% | estimated ± 3.5 pp, high confidence |
| Grok 4.1 Fast | 17.2% | 84.0% | estimated ± 3.5 pp, high confidence |
| Grok 4.1 Fast (Reasoning) | 25.1% | 85.6% | estimated ± 3.5 pp, high confidence |
| Grok 4.5 | 51.6% | 85.7% | estimated ± 3.5 pp, medium confidence |
| Grok 4.6 | 48.2% | 85.7% | estimated ± 3.5 pp, high confidence |
| Grok 4.7 | 47.4% | 85.7% | estimated ± 3.5 pp, high confidence |
| Grok 4 Fast (Reasoning) | 22.8% | 85.4% | estimated ± 3.5 pp, high confidence |
| Grok Code Fast 1 | 23.5% | 85.5% | estimated ± 3.5 pp, high confidence |
| Hy3 | 32.0% | 85.7% | estimated ± 3.5 pp, high confidence |
| K-Exaone | 16.4% | 83.6% | estimated ± 3.5 pp, high confidence |
| Kimi K2 | 27.4% | 85.6% | estimated ± 3.5 pp, high confidence |
| Kimi K2.7 Code | 39.6% | 85.7% | estimated ± 3.5 pp, high confidence |
| LFM2.5-2.6B | 4.4% | 48.9% | estimated ± 3.5 pp, medium confidence |
| LFM2.5-8B-A1B | 9.4% | 72.4% | estimated ± 3.5 pp, high confidence |
| LFM2.5-VL-1.6B-Extract | 5.8% | 56.9% | estimated ± 3.5 pp, medium confidence |
| Ling 3.0 Flash VL | 14.4% | 82.1% | estimated ± 3.5 pp, high confidence |
| Ling 3.0 Tiny | 8.5% | 69.3% | estimated ± 3.5 pp, high confidence |
| Ling 3.1 Flash | 29.1% | 85.7% | estimated ± 3.5 pp, high confidence |
| Llama 3.1 405B | 23.2% | 85.4% | estimated ± 3.5 pp, high confidence |
| Llama 4 Maverick | 24.9% | 85.5% | estimated ± 3.5 pp, high confidence |
| Llama 4 Scout | 15.2% | 82.8% | estimated ± 3.5 pp, high confidence |
| Mercury 2.5 | 22.0% | 85.3% | estimated ± 3.5 pp, high confidence |
| MiMo-V2.6-Flash | 27.0% | 85.6% | estimated ± 3.5 pp, high confidence |
| MiMo-V2.6-Pro | 34.8% | 85.7% | estimated ± 3.5 pp, high confidence |
| MiMo-V2-Omni | 19.3% | 84.8% | estimated ± 3.5 pp, high confidence |
| MiMo-V2-Pro | 26.6% | 85.6% | estimated ± 3.5 pp, high confidence |
| MiniMax M2.7 | 26.8% | 85.6% | estimated ± 3.5 pp, high confidence |
| MiniMax M3 | 16.7% | 83.8% | estimated ± 3.5 pp, high confidence |
| Mistral Large 2 | 19.9% | 84.9% | estimated ± 3.5 pp, high confidence |
| Mistral Large 3 | 25.0% | 85.6% | estimated ± 3.5 pp, high confidence |
| Mistral Large 4 | 25.8% | 85.6% | estimated ± 3.5 pp, high confidence |
| Mistral Medium 3 | 18.3% | 84.5% | estimated ± 3.5 pp, high confidence |
| Mistral Medium 3.5 128B | 24.7% | 85.5% | estimated ± 3.5 pp, high confidence |
| Mistral Small 4 | 21.7% | 85.3% | estimated ± 3.5 pp, high confidence |
| Mistral Small 4 (Reasoning) | 21.7% | 85.3% | estimated ± 3.5 pp, high confidence |
| Muse Glimmer 30B | 27.0% | 85.6% | estimated ± 3.5 pp, high confidence |
| Muse Spark 1.2 | 45.4% | 85.7% | estimated ± 3.5 pp, high confidence |
| Muse Spark 1.3 | 43.6% | 85.7% | estimated ± 3.5 pp, high confidence |
| Nemotron 3 Nano 30B | 17.3% | 84.1% | estimated ± 3.5 pp, high confidence |
| Nemotron 3 Super 100B | 24.3% | 85.5% | estimated ± 3.5 pp, high confidence |
| Nemotron Ultra 253B | 20.1% | 85.0% | estimated ± 3.5 pp, high confidence |
| North Mini Code | 18.9% | 84.7% | estimated ± 3.5 pp, high confidence |
| Nova Pro | 16.9% | 83.9% | estimated ± 3.5 pp, high confidence |
| o3 | 38.6% | 85.7% | estimated ± 3.5 pp, high confidence |
| Phi-4 | 14.1% | 81.8% | estimated ± 3.5 pp, high confidence |
| Quasar 438B | 15.5% | 83.0% | estimated ± 3.5 pp, high confidence |
| Qwen3.5 397B (Reasoning) | 24.5% | 85.5% | estimated ± 3.5 pp, high confidence |
| Qwen 3.6 Max (preview) | 37.9% | 85.7% | estimated ± 3.5 pp, high confidence |
| Qwen3.8 Max Preview | 31.7% | 85.7% | estimated ± 3.5 pp, high confidence |
| Qwen3 Max | 24.4% | 85.5% | estimated ± 3.5 pp, high confidence |
| Qwen3-Omni-30B-A3B-Instruct | 14.3% | 82.0% | estimated ± 3.5 pp, high confidence |
| Qwen3-Omni-30B-A3B-Thinking | 14.6% | 82.3% | estimated ± 3.5 pp, high confidence |
| Sarvam 105B | 17.6% | 84.2% | estimated ± 3.5 pp, high confidence |
| Sarvam 30B | 12.6% | 79.8% | estimated ± 3.5 pp, high confidence |
| Solar Pro 2 | 16.1% | 83.4% | estimated ± 3.5 pp, high confidence |
| Solar Pro 3 | 18.5% | 84.6% | estimated ± 3.5 pp, high confidence |
| Step 3.7 Flash | 25.8% | 85.6% | estimated ± 3.5 pp, high confidence |
| Step 5 Preview | 41.5% | 85.7% | estimated ± 3.5 pp, high confidence |
| Trinity-Large-Preview | 22.5% | 85.4% | estimated ± 3.5 pp, high confidence |
| Trinity-Large-Thinking | 22.5% | 85.4% | estimated ± 3.5 pp, high confidence |
| Ultravox v0.6 Llama 3.3 70B | 19.0% | 84.7% | estimated ± 3.5 pp, high confidence |