whatmodel.sh / MedGemma 4B IT

MedGemma 4B IT

+ contribute
Update model informationAdd a place to run itAdd a test resultPropose a benchmarkCorrect a benchmark scoreEdit data on GitHub

MedGemma 4B IT is an open-weights language model from Google, released May 2025. Its published size is 4.3B parameters. Licence: health-ai-developer-foundations.

Catalogue identity notes

google/medgemma-4b-it @290cda5 (created 2025-05-19, licence health-ai-developer-foundations). Google MedGemma (image-text-to-text; SigLIP medical image encoder.) built on Gemma 3; gated under the Health AI Developer Foundations terms. Independent rows: BRIDGE clinical zero-shot (87 tasks) and Open PT-LLM.

Model information sources

Release date ↗ · Weights licence ↗

HF safetensors metadata 4,300,079,472 parameters.

46.4±2.7#413 of 719 LLMsfrom 4 benchmark results across 4 sources

$ how it compares

The LLMs ranked either side of MedGemma 4B IT, out of 828 with a score, then more from Google.

#modelscorewhereRAMreleased
490Mistral Large (2402) Mistral46±1comparison only2024-02
491Jamba 1.5 Mini AI21 Labs46±1local~34GB2024-08
492MedGemma 4B IT Google46±3local~4.1GB2025-05
493OLMo 2 13B Instruct (1124) Allen Institute for AI46±1local~9.6GB2024-12
494Falcon-H1 7B Instruct TII46±3local~6GB2025-05
more from Google
9Gemini 4 Argon (limited preview) Google82±1limited preview2026-09
23Gemini 3.8 Flash Google77±1api $1.50/M2026-09
32Gemini 3.7 Flash Google76±1api $1.50/M2026-08

see all LLMs · find the ones that fit your machine

$ what it's good at

other tests4 not in a task · 97% of its score
testresultvs the others in its rundifficultycountswhy
Polish medical board exams, bias-mitigated MCQ protocol (7-exam average)43.4% ✎▄·▂▂▃··█▂▄·▄ beats 4/13hard · avg 48%▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇ 37%at its level
BRIDGE clinical text, zero-shot average (87 tasks)29.4% ✎▁▁▁▁▁▂▂▃▄▆█▃ beats 29/97hard · avg 32%▇▇▇▇▇▇▇▇▇▇▇▇ 31%blunt · at its level
Open PT-LLM Leaderboard average (9 Portuguese tasks)68.2% ✎▃▁▁▁▁▁▁▁▂█▅▁ beats 149/276medium · avg 73%▇▇▇▇▇▇ 15%at its level · noisy
SurgiQ surgical MCQ (zero-shot, log-probability scoring)44.1% ✎▂···▂▂·▃▂▃██ beats 4/17medium · avg 53%▇▇▇▇▇▇ 14%at its level

$ where to run it

llama.cpp · recommended

version

  1. Install llama.cpp · Homebrew and winget installs name the commands llama-cli and llama-server instead.
    macOS / Linux
    curl -LsSf https://llama.app/install.sh | sh
    Windows
    irm https://llama.app/install.ps1 | iex

    other ways to install ↗

  2. Run MedGemma 4B IT · needs ~2.5 GB · from huggingface.co/lmstudio-community
    chat in the terminal
    llama cli -hf lmstudio-community/medgemma-4b-it-GGUF:Q4_K_M
  3. Or start a local server and open the address it prints (http://localhost:9931)
    llama serve -hf lmstudio-community/medgemma-4b-it-GGUF:Q4_K_M

LM Studio

version

  1. Install LM Studio · Open the app once before using the lms command.

    download it (macOS / Windows / Linux) ↗

  2. Run MedGemma 4B IT · needs ~2.5 GB · from huggingface.co/lmstudio-community

    open in LM Studio ↗

  3. Or download it in the terminal
    lms get https://huggingface.co/lmstudio-community/medgemma-4b-it-GGUF@Q4_K_M -y
  4. Then run lms ls to see its name, and lms chat followed by that name

Ollama

  1. Install Ollama

    download it (macOS / Windows) ↗

    Linux
    curl -fsSL https://ollama.com/install.sh | sh
  2. Run MedGemma 4B IT · needs ~3.3 GB · from ollama.com
    chat in the terminal
    ollama run medgemma:4b

MLX (Mac)

  1. Install MLX (Mac) · Apple silicon Macs only, text only.
    with uv
    uv tool install mlx-lm
    or
    pip install mlx-lm
  2. Run MedGemma 4B IT · needs ~3 GB, Mac only · from huggingface.co/lmstudio-community
    chat in the terminal
    mlx_lm.chat --model lmstudio-community/medgemma-4b-it-MLX-4bit
  3. Or start an OpenAI-compatible server at http://localhost:8080/v1
    mlx_lm.server --model lmstudio-community/medgemma-4b-it-MLX-4bit

Official weights and documentation: huggingface.co/google ↗

$ how the score was worked out · Reassessing High-Performing L… 37%, BRIDGE Medical Leaderboard (H… 31%, Open Portuguese LLM Leaderboa… 15% and 1 more

Each result implies a score on its own, after adjusting for how hard its test is and who ran it. The score is their weighted average, with a small pull toward the middle for thinly tested models. Each test's weight is its counts column above; here it's added up by who ran the tests. How it works

sourceresultsshare of the scorelicence
Reassessing High-Performing LLMs on Polish Medical Exams: True Competence or Bias-Driven Performance? (arXiv:2606.12250v1)1▇▇▇▇▇▇▇▇▇▇▇ 37%CC BY-SA 4.0
BRIDGE Medical Leaderboard (Harvard Medical School / Brigham and Women's YLab)1▇▇▇▇▇▇▇▇▇ 31%MIT
Open Portuguese LLM Leaderboard (archived)1▇▇▇▇ 15%Apache-2.0
SurgiQ: A Large-Scale Multi-Domain Benchmark for Evaluating Surgical Understanding in Large Language Models (arXiv:2606.08071v1)1▇▇▇▇ 14%CC BY 4.0
prior · pull toward the middle▇ 3%
weighted average= 46.4

$ improve this page

Update model informationAdd a place to run itAdd a test resultPropose a benchmarkCorrect a benchmark scoreEdit data on GitHub

Use a short GitHub form or edit the data through a pull request. Every number needs a source; here's which ones we can use.

data updated · sources and licences · json