MedQA
US medical licensing exam questions. Run by Vals AI.
- Published by
- Vals AI
- Category
- Knowledge
- Index weight
- 0.5
- Models
- 90
- Data as of
- 9 Sept 2026
Vals AI (vals.ai).
- 1o196.5%
- 2GPT-5.196.4%
- 3Gemini 3.1 Pro96.4%
- 4GPT-596.3%
- 5GPT-5.496.1%
- 6o396.1%
- 7GPT-5 mini96.1%
- 8Gemini 3 Pro96.0%
- 9o4-mini96.0%
- 10Claude Opus 4.595.9%
- 11Gemini 3 Flash95.8%
- 12Claude Opus 4.695.4%
- 13Qwen3.5 Plus95.2%
- 14o3-mini94.8%
- 15Claude Sonnet 4.594.7%
90 of 90
| # | |||||
|---|---|---|---|---|---|
| 1 |  o1OpenAI | 96.5% | – | 44.3 | – |
| 2 | GPT-5.1OpenAI | 96.4% | – | 53.8 | – |
| 3 |  Gemini 3.1 ProGoogle | 96.4% | – | 60.7 | – |
| 4 | GPT-5OpenAI | 96.3% | – | 54.5 | – |
| 5 | GPT-5.4OpenAI | 96.1% | – | 61.3 | – |
| 6 | o3OpenAI | 96.1% | – | 52.9 | – |
| 7 | GPT-5 miniOpenAI | 96.1% | – | 52.3 | – |
| 8 | Gemini 3 ProGoogle | 96.0% | – | 57.0 | – |
| 9 | o4-miniOpenAI | 96.0% | – | 48.0 | – |
| 10 |  Claude Opus 4.5Anthropic | 95.9% | – | 60.6 | – |
| 11 | Gemini 3 FlashGoogle | 95.8% | – | 58.5 | – |
| 12 | Claude Opus 4.6Anthropic | 95.4% | – | 62.3 | – |
| 13 |  Qwen3.5 PlusAlibaba | 95.2% | – | 57.2 | – |
| 14 | o3-miniOpenAI | 94.8% | – | 47.8 | – |
| 15 | Claude Sonnet 4.5Anthropic | 94.7% | – | 52.8 | – |
| 16 |  Grok 4.20xAI | 94.5% | – | 54.3 | – |
| 17 |  Kimi K2.5Moonshot AIopen | 94.4% | – | 55.3 | – |
| 18 |  GLM-5Zhipu AIopen | 94.3% | – | 57.0 | – |
| 19 | GPT-5.2OpenAI | 94.1% | – | 57.3 | – |
| 20 |  DeepSeek V3.2DeepSeekopen | 93.9% | – | 50.1 | – |
| 21 | GLM-4.7Zhipu AIopen | 93.7% | zai/glm-4.7 | 54.9 | – |
| 22 | Claude Opus 4.1Anthropic | 93.6% | – | 58.1 | – |
| 23 | GPT-5 nanoOpenAI | 93.3% | – | 45.9 | – |
| 24 | Claude Opus 4.5Anthropic | 93.2% | – | 58.0 | – |
| 25 | Gemini 2.5 ProGoogle | 93.1% | – | 54.1 | – |
| 26 | Claude Opus 4Anthropic | 92.9% | – | 50.7 | – |
| 27 | Claude Sonnet 4Anthropic | 92.7% | – | 54.0 | – |
| 28 | Kimi K2 ThinkingMoonshot AIopen | 92.6% | – | 52.8 | – |
| 29 |  MiniMax-M2.5MiniMaxopen | 92.5% | minimax/MiniMax-M2.5 | 53.5 | – |
| 30 | Claude Opus 4.1Anthropic | 92.5% | – | 52.6 | – |
| 31 | Grok 4xAI | 92.5% | 0709 | 57.2 | – |
| 32 | Grok 2xAI | 92.3% | 1212 | 41.6 | – |
| 33 | GLM-4.6Zhipu AIopen | 92.2% | zai/glm-4.6 | 47.6 | – |
| 34 | Grok 4.1xAI | 92.1% | – | 55.9 | – |
| 35 | Grok 4 FastxAI | 92.1% | – | 53.0 | – |
| 36 | Claude Sonnet 4.6Anthropic | 92.1% | – | 60.3 | – |
| 37 | Gemini 2.5 Flash 09Google | 91.4% | preview, 2025 | 50.7 | – |
| 38 | gpt-oss-120bOpenAIopen | 91.4% | – | 47.6 | – |
| 39 | GPT-4.1OpenAI | 91.2% | – | 46.2 | – |
| 40 | Gemini 2.5 Flash 09Google | 91.2% | thinking, preview, 2025 | 52.5 | – |
| 41 | MiniMax-M2MiniMaxopen | 91.2% | minimax/MiniMax-M2.1 | 55.1 | – |
| 42 | Gemini 2.5 FlashGoogle | 91.0% | – | 48.1 | – |
| 43 | DeepSeek R1DeepSeekopen | 90.8% | fireworks/deepseek-r1 | 48.9 | – |
| 44 | Qwen3 235B A22BAlibabaopen | 90.6% | – | 48.7 | – |
| 45 | Claude Sonnet 4Anthropic | 90.3% | – | 49.5 | – |
| 46 | Claude 3.7 SonnetAnthropic | 90.2% | – | 52.1 | – |
| 47 | o1-miniOpenAI | 90.2% | – | – | – |
| 48 | Grok 3 Mini FastxAI | 90.1% | – | 53.5 | – |
| 49 | GLM-4.5Zhipu AIopen | 90.0% | zai/glm-4.5 | 50.2 | – |
| 50 |  Magistral MediumMistral AIopen | 89.5% | medium, 2509 | 44.1 | – |
| 51 | Gemini 2.5 Flash Lite 09Google | 88.9% | – | 48.8 | – |
| 52 | Grok 3 Mini FastxAI | 88.7% | – | 50.3 | – |
| 53 |  Llama Meta Llama 3.1 405B TurboMeta | 88.2% | – | – | – |
| 54 | GPT-4oOpenAI | 88.2% | – | 41.1 | – |
| 55 | Qwen3 MaxAlibaba | 87.4% | – | 51.8 | – |
| 56 | Qwen3 MaxAlibaba | 87.4% | – | 52.7 | – |
| 57 | Gemini 2.5 FlashGoogle | 86.7% | – | 48.8 | – |
| 58 | Llama Meta Llama 3.1 70B TurboMeta | 84.8% | – | – | – |
| 59 | GPT-4.1 miniOpenAI | 84.6% | – | 45.9 | – |
| 60 | Kimi K2Moonshot AIopen | 84.0% | – | 48.8 | – |