MedGemma App

Streamlit app for medical image and text inference using Google's MedGemma model in MLX format, running locally on Mac with Apple Silicon.

Features

• Single-column UI with text input and optional image upload
• Collapsible settings for thinking trace and system instruction
• System instruction auto-adjusts based on whether an image is attached
• Fully local inference on Apple Silicon via MLX

Setup

Requires:

• Mac with Apple Silicon
• Python 3.12
• uv
• Hugging Face token with access to google/medgemma-1.5-4b-it

uv sync

Create a .env file:

HF_TOKEN=your_token_here

Usage

uv run streamlit run streamlit_app.py

Development

uv run ruff check .               # Lint
uv run ruff format .              # Format
uv run ty check                   # Type check
uv run pytest                     # Run tests