MMLU subject subset: General clinical medicine facts and patient-care knowledge, at a level below the licensing-exam-style Professional Medicine subject.
unassessed
| Category | knowledge |
|---|---|
| Subcategory | health |
| Page status | active |
| Metric | accuracy |
| Direction | higher_is_better |
| Unit | % |
| Dataset size | 265 |
| Dataset licence | MIT |
| Publisher | UC Berkeley (original); Center for AI Safety (current host) |
General clinical medicine facts and patient-care knowledge, at a level below the licensing-exam-style Professional Medicine subject. Questions are four-option multiple-choice, drawn from the MMLU test set's "health" subcategory within the benchmark's "other" top-level group, and are graded on the single correct labelled option.
Four-option multiple-choice questions, graded on the single correct labelled option; commonly evaluated 5-shot, consistent with the rest of MMLU.
| Model | Provider | Score | Card as of |
|---|---|---|---|
| Claude Opus 4 | Anthropic | 87.5 | 2026-04 |
| Claude Opus 4.6 | Anthropic | 87.5 | 2026-04 |
| Meta Llama 3 70B | Meta | 86.8 | 2024-07 |
| Meta Llama 3 70B Instruct | Meta | 86.8 | 2026-04 |
| Meta Llama 3 70B Instruct | Nous Research | 86.8 | 2026-04 |
| Gemini 2.5 Pro | Google DeepMind | 86.5 | 2026-04 |
| GPT-4.1 | OpenAI | 86.2 | 2026-04 |
| GPT-4o | OpenAI | 85.5 | 2026-04 |
| GPT-4o (2024-05-13) | OpenAI | 85.5 | 2026-04 |
| GPT-4o (2024-08-06) | OpenAI | 85.5 | 2026-04 |
| GPT-4o (2024-11-20) | OpenAI | 85.5 | 2026-04 |
| GPT-4o mini | OpenAI | 85.5 | 2026-04 |
| Claude Sonnet 4 | Anthropic | 84.2 | 2026-04 |
| Claude Sonnet 4.5 | Anthropic | 84.2 | 2026-04 |
| Claude Sonnet 4.5 (latest) | Anthropic | 84.2 | 2026-04 |
| DeepSeek R1 | DeepSeek | 83.5 | 2026-04 |
| DeepSeek R1 0528 | DeepSeek | 83.5 | 2026-04 |
| DeepSeek R1 0528 NVFP4 v2 | NVIDIA | 83.5 | 2026-04 |
| DeepSeek R1 0528 Qwen3 8B | DeepSeek | 83.5 | 2026-04 |
| DeepSeek R1 Distill Llama 70B | DeepSeek | 83.5 | 2026-04 |
| DeepSeek R1 Distill Llama 8B | DeepSeek | 83.5 | 2026-04 |
| DeepSeek R1 Distill Qwen 1.5B | DeepSeek | 83.5 | 2026-04 |
| DeepSeek R1 Distill Qwen 14B | DeepSeek | 83.5 | 2026-04 |
| DeepSeek R1 Distill Qwen 32B | DeepSeek | 83.5 | 2026-04 |
| DeepSeek R1 Distill Qwen 7B | DeepSeek | 83.5 | 2026-04 |
| DeepSeek Reasoner | DeepSeek | 83.5 | 2026-04 |
| Yi 1.5 34B | 01.AI | 83.0 | 2026-04 |
| Yi 1.5 34B 32K | 01.AI | 82.6 | 2024-07 |
| Mixtral 8x22B Instruct v0.1 | Mistral AI | 82.3 | 2026-04 |
| Qwen 3 235B Instruct | Cerebras | 82.2 | 2026-04 |
| Qwen3 235B-A22B | Alibaba / Qwen Team | 82.2 | 2026-04 |
| Yi 34B 200K | 01.AI | 81.1 | 2024-07 |
| Mistral Large (latest) | Mistral AI | 80.5 | 2026-04 |
| Mistral Large 2.1 | Mistral AI | 80.5 | 2026-04 |
| Mistral Large 3 | Mistral AI | 80.5 | 2026-04 |
| Gemma 4 31B | Google DeepMind | 80.2 | 2026-04 |
| gemma 4 31B it | Google DeepMind | 80.2 | 2026-04 |
| gemma 4 31B it GGUF | Unsloth | 80.2 | 2026-04 |
| Gemma 4 31B IT NVFP4 | NVIDIA | 80.2 | 2026-04 |
| Nous Hermes 2 Yi 34B | Nous Research | 80.0 | 2024-07 |
| Gemma 4 26B | Google DeepMind | 79.2 | 2026-04 |
| Nous Hermes 2 Mixtral 8x7B DPO | Nous Research | 79.2 | 2024-07 |
| Yi 34B Chat | 01.AI | 79.2 | 2024-07 |
| Llama 3.3 70B Instruct NVFP4 | NVIDIA | 78.5 | 2026-04 |
| Llama-3.3-70B-Instruct | Meta | 78.5 | 2026-04 |
| Mixtral 8x7B v0.1 | Mistral AI | 78.5 | 2026-04 |
| Mixtral 8x7B Instruct v0.1 | Mistral AI | 77.7 | 2026-04 |
| Yi 1.5 34B Chat | 01.AI | 77.4 | 2026-04 |
| Yi 1.5 34B Chat 16K | 01.AI | 77.4 | 2024-07 |
| Llama 3.1 70B | Meta | 77.2 | 2026-04 |
| Llama 3.1 70B Instruct | Meta | 77.2 | 2026-04 |
| phi 4 | Microsoft | 76.2 | 2026-04 |
| Phi 4 mini instruct | Microsoft | 76.2 | 2026-04 |
| Phi 4 multimodal instruct | Microsoft | 76.2 | 2026-04 |
| Meta Llama 3 8B | Meta | 75.5 | 2024-07 |
| Meta Llama 3 8B | Nous Research | 75.5 | 2024-07 |
| Meta Llama 3 8B Instruct | Meta | 74.7 | 2024-07 |
| Meta Llama 3 8B Instruct | Nous Research | 74.7 | 2024-07 |
| Hermes 2 Theta Llama 3 8B | Nous Research | 74.3 | 2024-07 |
| Phi 3 mini 4K instruct | Microsoft | 74.3 | 2024-07 |
| Phi 3 mini 128K instruct | Microsoft | 74.0 | 2024-07 |
| Hermes 2 Pro Llama 3 8B | Nous Research | 73.6 | 2024-07 |
| Yi 9B | 01.AI | 73.2 | 2024-07 |
| Yi 1.5 9B Chat | 01.AI | 72.8 | 2024-07 |
| Yi 1.5 9B Chat 16K | 01.AI | 72.5 | 2024-07 |
| Yi 1.5 9B 32K | 01.AI | 71.7 | 2024-07 |
| Nous Hermes 2 SOLAR 10.7B | Nous Research | 69.1 | 2024-07 |
| gemma 7B it | Google DeepMind | 68.7 | 2024-07 |
| Mistral 7B v0.3 | Mistral AI | 68.3 | 2024-07 |
| mistral 7B v0.3 bnb 4bit | Unsloth | 68.3 | 2024-07 |
| Yi 1.5 9B | 01.AI | 68.3 | 2024-07 |
| Yi 1.5 6B Chat | 01.AI | 67.9 | 2024-07 |
| Mistral 7B Instruct v0.2 | Mistral AI | 67.2 | 2024-07 |
| Yi 6B | 01.AI | 66.8 | 2024-07 |
| Yi 6B Chat | 01.AI | 66.8 | 2024-07 |
| Yi 1.5 6B | 01.AI | 65.7 | 2024-07 |
| falcon 40B | TII | 60.8 | 2024-07 |
| phi 2 | Microsoft | 60.4 | 2024-07 |
| Qwen2 1.5B Instruct | Alibaba / Qwen Team | 60.0 | 2024-07 |
| deepseek llm 7B base | DeepSeek | 53.2 | 2024-07 |
| deepseek llm 7B chat | DeepSeek | 53.2 | 2024-07 |
| Qwen2 0.5B Instruct | Alibaba / Qwen Team | 50.2 | 2024-07 |
| chatglm2 6B | Zhipu AI | 48.7 | 2024-07 |
| gemma 2B | Google DeepMind | 46.8 | 2024-07 |
| gemma 2B it | Google DeepMind | 42.3 | 2024-07 |
| deepseek coder 6.7B base | DeepSeek | 41.1 | 2024-07 |
| deepseek coder 6.7B instruct | DeepSeek | 41.1 | 2024-07 |
| deepseek coder 1.3B base | DeepSeek | 30.2 | 2024-07 |
| deepseek coder 1.3B instruct | DeepSeek | 30.2 | 2024-07 |
| OLMo 1B hf | Allen AI | 19.6 | 2024-07 |