Skip to main contentThe Local Bench

Category ranking · 8 models · checked 2026-10-13

Biomedical

Ranked by MedQA (4-option), accuracy. Source: Google MedGemma model card.

  1. MedGemma 4B64.4
  2. MedGemma 1.5 4B69.1
  3. MedGemma 27B85.3
MedQA (4-option), accuracy. Longer is better. Values are exactly as published; the table below links each one.

MedQA (4-option), accuracy

3 with a published score, 5 not yet measured.

#
Model
Parameters
Score
Source
Licence
1MedGemma 27B
Google
28.8 B85.3%SourcedSource
"MedGemma 1 27B" column (multimodal variant per the page), text evaluations table.
Health AI Developer Foundations terms
2MedGemma 1.5 4B
Google
4.3 B69.1%SourcedSource
"MedGemma 1.5 4B" column, text evaluations table.
Health AI Developer Foundations terms
3MedGemma 4B
Google
4.3 B64.4%SourcedSource
"MedGemma 1 4B" column, text evaluations table.
Health AI Developer Foundations terms
—II-Medical 8B
Intelligent Internet
8.2 BNot yet measured—Apache-2.0
—Med42 8B
M42
8.0 BNot yet measured—Llama 3 Community
—MedGemma 27B text
Google
27.0 BNot yet measuredNot in the current model-card table.Health AI Developer Foundations terms
—OpenBioLLM 8B
Saama AI (aaditya)
—Not yet measured—Llama 3 Community
—TxGemma 9B chat
Google
9.2 BNot yet measured—Health AI Developer Foundations terms

How to read this: one public benchmark per category, so you can compare like with like. We copy the published number and link it; we do not re-run it, average different benchmarks or fill gaps. Vendor-reported numbers use each vendor's own settings, so treat small gaps as ties. “Not yet measured” means we found no comparable public score, not that the model is weak.

Every model was checked against the Hugging Face API on 2026-10-13: it exists, is public, and has the parameter count and licence shown. Other categories: Math and reasoning · Coding · Vision-language · OCR and documents · Speech to text · Text to speech · Image generation · Embeddings and rerankers · Small fast 'System 1' · Tool use and agents · Multilingual and translation · Time series · Robotics (VLA).

Machine-readable: /api/v1/categories/biomedical.json