Open Medical LLM - PubMedQA: leaderboard
PubMedQA subset: biomedical yes/no/maybe QA requiring reasoning over PubMed research article abstracts.
Metric: Accuracy (%). Source: huggingface.co. Saturation forecast: Estimated already saturated. 193 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Llama 3 8B Ita | 78.4 | #926 |
| 2 | Llama-3-SauerkrautLM-8B-Instruct | 78.4 | #895 |
| 3 | Starling-LM-7B-beta | 78.2 | #1093 |
| 4 | llama-3-merged-linear | 78.2 | #887 |
| 5 | Myrrh_solar_10.7b_3.0 | 78 | #1042 |
| 6 | Barcenas-Llama3-8B-ORPO | 77.6 | #949 |
| 7 | NeuralLLaMa-3-8B-DT-v0.1 | 77.6 | #923 |
| 8 | Daredevil-8B-abliterated-dpomix | 77.4 | #838 |
| 9 | Daredevil-8B-abliterated | 77 | #846 |
| 10 | NeuralLLaMa-3-8B-ORPO-v0.3 | 77 | #1174 |
| 11 | zephyr-7B-beta | 76.6 | #1330 |
| 12 | L-MChat-7B | 76.6 | #1120 |
| 13 | ChimeraLlama-3-8B-v3 | 76.4 | #1055 |
| 14 | Master-Yi-9B | 76.2 | #1119 |
| 15 | Llama-3-8B-Instruct-abliterated-dpomix | 76.2 | #1049 |
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-pubmedqa · How It Works · Data refreshed daily, snapshot 2026-10-20.