Open Medical LLM: leaderboard
Medical question-answering accuracy across MedQA, MedMCQA, PubMedQA and six medical MMLU subjects. The aggregate requires all nine tasks. Submitted numerical precision and checkpoint revisions remain distinct; the source describes zero-shot evaluation with separately attributed paper-import exceptions.
Metric: Average Accuracy (%). Source: huggingface.co. Saturation forecast: Estimated already saturated. 193 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | GPT-4 | 82.97 | #438 |
| 2 | Daredevil-8B-abliterated-dpomix | 70.99 | #838 |
| 3 | Daredevil-8B-abliterated | 70.86 | #846 |
| 4 | Llama 3 8B Slerp Biomed Chat Chinese | 70.78 | #1215 |
| 5 | NeuralLLaMa-3-8B-DT-v0.1 | 70.43 | #923 |
| 6 | NeuralLLaMa-3-8B-ORPO-v0.3 | 70.4 | #1174 |
| 7 | llama-3-merged-linear | 70.13 | #887 |
| 8 | Llama 3 8B Ita | 70.11 | #926 |
| 9 | ChimeraLlama-3-8B-v3 | 70.05 | #1055 |
| 10 | Barcenas-Llama3-8B-ORPO | 69.93 | #949 |
| 11 | Llama-3-SauerkrautLM-8B-Instruct | 69.76 | #895 |
| 12 | Llama-3-8B-Instruct-abliterated-dpomix | 69.65 | #1049 |
| 13 | suzume-llama-3-8B-multilingual | 69.3 | #1084 |
| 14 | Master-Yi-9B | 69.07 | #1119 |
| 15 | Llama 3 8B Instruct | 68.99 | #1129 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=open-medical-llm · How It Works · Data refreshed daily, snapshot 2026-10-20.