Calibration
GPQA Diamond → MedXpertQA (Text)
MedXpertQA (Text) is estimated from GPQA Diamond with a inverse Michaelis–Menten curve fitted on 5 models measured on both: y = 0.13678·(x − 0.0000) / (0.0000 + 1.1290 − x), R² = 0.94, cross-validated error 4.5 pp. It is used for 62 estimates.
| Estimated model | GPQA Diamond | MedXpertQA (Text) | Source |
|---|---|---|---|
| A.X K2 | 85.6% | 42.9% | estimated ± 4.5 pp, low confidence |
| Claude Opus 4.7 (Adaptive) | 94.2% | 68.9% | estimated ± 4.5 pp, medium confidence |
| Claude Opus 4.8 | 93.6% | 66.3% | estimated ± 4.5 pp, medium confidence |
| DeepSeek V4.1 Flash | 90.9% | 56.5% | estimated ± 4.5 pp, medium confidence |
| DeepSeek V4 Flash 0731 | 88.1% | 48.6% | estimated ± 4.5 pp, low confidence |
| DeepSeek V4 Pro 0813 | 90.1% | 54.1% | estimated ± 4.5 pp, medium confidence |
| Gemini 3.5 Flash | 92.7% | 62.7% | estimated ± 4.5 pp, medium confidence |
| Gemma 4 12B | 78.8% | 31.6% | estimated ± 4.5 pp, low confidence |
| GLM-5 | 86.0% | 43.7% | estimated ± 4.5 pp, low confidence |
| GLM-5.1 | 86.2% | 44.2% | estimated ± 4.5 pp, low confidence |
| GLM-5.2 | 91.2% | 57.5% | estimated ± 4.5 pp, medium confidence |
| GPT-5.5 | 93.6% | 66.3% | estimated ± 4.5 pp, medium confidence |
| GPT-5.6 Luna | 92.3% | 61.3% | estimated ± 4.5 pp, medium confidence |
| GPT-5.6 Sol | 94.6% | 70.7% | estimated ± 4.5 pp, low confidence |
| GPT-5.6 Terra | 92.9% | 63.5% | estimated ± 4.5 pp, medium confidence |
| GPT-6 Astra | 96.0% | 77.7% | estimated ± 4.5 pp, low confidence |
| Hy3 Preview | 87.2% | 46.4% | estimated ± 4.5 pp, low confidence |
| Hy4 preview | 92.3% | 61.3% | estimated ± 4.5 pp, medium confidence |
| Inkling | 87.9% | 48.1% | estimated ± 4.5 pp, low confidence |
| Inkling-Small | 89.5% | 52.3% | estimated ± 4.5 pp, medium confidence |
| Interfaze Beta | 89.9% | 53.5% | estimated ± 4.5 pp, medium confidence |
| K-EXAONE 2.0 | 82.2% | 36.6% | estimated ± 4.5 pp, low confidence |
| Kimi K2.6 | 90.5% | 55.3% | estimated ± 4.5 pp, medium confidence |
| Kimi K2.5 | 87.6% | 47.4% | estimated ± 4.5 pp, low confidence |
| Kimi K3 | 93.5% | 65.9% | estimated ± 4.5 pp, medium confidence |
| LFM2.5-230M | 25.4% | 4.0% | estimated ± 4.5 pp, low confidence |
| Ling 3.0 Flash | 85.0% | 41.6% | estimated ± 4.5 pp, low confidence |
| Ling 3.0 Flash FP8 | 84.0% | 39.8% | estimated ± 4.5 pp, low confidence |
| LLaDA2.2-mini | 44.4% | 8.9% | estimated ± 4.5 pp, low confidence |
| LongCat-Flash-Lite-Sparse | 69.5% | 21.9% | estimated ± 4.5 pp, low confidence |
| MAI-Thinking-1 | 84.2% | 40.1% | estimated ± 4.5 pp, low confidence |
| Mellum2-12B-A2.5B-Instruct | 40.9% | 7.8% | estimated ± 4.5 pp, low confidence |
| Mellum2-12B-A2.5B-Thinking | 57.6% | 14.2% | estimated ± 4.5 pp, low confidence |
| Mercury 2.5 | 79.0% | 31.9% | estimated ± 4.5 pp, low confidence |
| MiniCPM5-1B | 26.3% | 4.1% | estimated ± 4.5 pp, low confidence |
| MiniCPM5-2B | 70.2% | 22.5% | estimated ± 4.5 pp, low confidence |
| MiniMax M2.7 | 87.0% | 45.9% | estimated ± 4.5 pp, low confidence |
| Nemotron 3.5 Lightning 30B A3B NVFP4 | 75.6% | 27.7% | estimated ± 4.5 pp, low confidence |
| Nemotron 3 Nano Omni 30B A3B | 72.2% | 24.3% | estimated ± 4.5 pp, low confidence |
| Nemotron 3 Ultra | 87.0% | 45.9% | estimated ± 4.5 pp, low confidence |
| Ornith-1.5-35B-A3B | 89.2% | 51.5% | estimated ± 4.5 pp, medium confidence |
| Ornith-1.5-397B | 92.8% | 63.2% | estimated ± 4.5 pp, medium confidence |
| Ornith-1.5-9B | 86.4% | 44.6% | estimated ± 4.5 pp, low confidence |
| Pareto 26.10 Preview | 92.4% | 61.7% | estimated ± 4.5 pp, medium confidence |
| Qwen3.7 Max | 92.4% | 61.7% | estimated ± 4.5 pp, medium confidence |
| Qwen3.7 Plus | 90.3% | 54.7% | estimated ± 4.5 pp, medium confidence |
| Qwen3.8-27B | 89.2% | 51.5% | estimated ± 4.5 pp, medium confidence |
| Qwen3.8-Flash-Next | 91.7% | 59.2% | estimated ± 4.5 pp, medium confidence |
| Qwen3.8 Max | 92.6% | 62.4% | estimated ± 4.5 pp, medium confidence |
| Qwen3.8-Omni-Flash | 91.0% | 56.8% | estimated ± 4.5 pp, medium confidence |
| Beam | 90.5% | 55.3% | estimated ± 4.5 pp, medium confidence |
| Sakana Fugu | 95.5% | 75.1% | estimated ± 4.5 pp, low confidence |
| Sakana Fugu-Ultra | 95.5% | 75.1% | estimated ± 4.5 pp, low confidence |
| Solar Open 2 | 86.3% | 44.4% | estimated ± 4.5 pp, low confidence |
| Solar Pro 4 | 89.0% | 50.9% | estimated ± 4.5 pp, medium confidence |
| Soofi S 30B-A3B | 43.4% | 8.5% | estimated ± 4.5 pp, low confidence |
| Step 5 Preview | 93.5% | 65.9% | estimated ± 4.5 pp, medium confidence |
| Ternary Bonsai 2 27B | 85.8% | 43.2% | estimated ± 4.5 pp, low confidence |
| Trinity-Large-Preview | 63.3% | 17.5% | estimated ± 4.5 pp, low confidence |
| Trinity-Large-Thinking | 76.3% | 28.5% | estimated ± 4.5 pp, low confidence |
| ZAYA1-74B-Preview | 57.3% | 14.1% | estimated ± 4.5 pp, low confidence |
| ZAYA1-8B | 71.0% | 23.2% | estimated ± 4.5 pp, low confidence |