MedMLX home AI-IR lab Get started

Medical imaging.
Native on Mac.

Run medical imaging models locally on Apple Silicon.

Browse models Read the article
uv tool install --python 3.13 "medmlx[all,mcp]"

Installs the medmlx command, its MCP server and every model family. Needs macOS 14.0 or later on Apple Silicon and uv.

Outputs from MedMLX runtimes. Image credits

MedMLX reimplements published medical imaging models in MLX, Apple’s machine learning framework. All inference runs on device.

Featured models

All models and tools

  • nnU-Net

    Segmentation of liver and pancreas CT, brain tumor MRI and PET/CT lesions.

  • TotalSegmentator

    Identify 117 organs and other body structures in CT scans.

  • MedSAM2

    Lesion or organ segmentation from a box, point or mask prompt, tracked through the volume.

  • BioMedParse

    Text-prompted 3D segmentation with Microsoft BioMedParse v2.

Models by task

Request a model

Intended use

MedMLX is research software and not a medical device. Outputs require human review, and matching the original model does not establish clinical validity.

About MedMLX