MedXpertQA MM (200-question set)
Multimodal
A 200-question set from MedXpertQA's multimodal split: expert-level five-option medical questions that pair clinical images with structured patient information across 11 body systems. Scores are answer accuracy.
Top models (higher is better)
| Model | Score |
|---|---|
| GPT-6 Astra | 86.8 |
| Claude Opus 5.5 | 85.7 |
| GPT-6.1 Sol | 85.7 |
| Gemini 3.8 Flash | 84.7 |
| Claude Fable 5.1 | 83.7 |
| Gemini 3.5 Flash | 83.5 |
| Gemini 3.7 Flash | 83.5 |
| Gemini 3.6 Flash | 82.8 |
| GPT-6 Sol | 81.3 |
| Claude Sonnet 5.5 | 80.7 |
| Gemini 3.1 Pro Preview | 80.2 |
| Claude Opus 5 | 79.8 |
| GPT-5.6 Sol Pro | 79.5 |
| Muse Spark 1.1 | 77.0 |
| GPT-5.5 | 76.7 |
| Muse Spark 1.3 | 76.3 |
| Claude Fable 5 | 76.2 |
| GPT-5.6 Terra | 75.7 |
| Muse Spark 1.2 | 75.7 |
| Grok 4.6 | 75.2 |
| GPT-5.4 | 74.2 |
| GLM-5.3 Flash | 73.7 |
| Kimi K3 | 73.2 |
| Opus 4.7 | 72.8 |
| Opus 4.8 | 72.5 |
| GPT-5.6 Luna | 70.2 |
| Grok 4.5 | 69.5 |
| Grok 4.7 | 69.0 |
| Claude Haiku 5.5 | 67.8 |
| Gemini 3.5 Flash-Lite | 66.7 |
| Qwen3.5 397B A17B | 65.8 |
| Kimi K2.7 Code | 65.7 |
| GPT-6 Luna | 64.8 |
| DeepSeek V4.1 Flash | 63.5 |
| Sonnet 5 | 63.0 |
| Sonnet 4.6 | 61.7 |
| Inkling | 60.0 |
| MiniMax M3 | 59.7 |
| GLM-5.3 | 49.2 |