spicyneuron/medgemma-1.5-4b-it-llamafile

MedGemma 1.5 4B Llamafile

Llamafile for google/medgemma-1.5-4b-it.

Note: Gemma3 multimodal isn't working in Llamafile 0.9.3. This is text-only for now.

Use

On Windows, add .exe to the llamafile and run as a program.

On Mac or Linux, open up a terminal and run:

# Start server at 127.0.0.1:8080
./medgemma.llamafile

# Or override the defaults
./medgemma.llamafile --port 9090 --n-gpu-layers 99

For more info:

• https://mozilla-ai.github.io/llamafile/quickstart/
• https://mozilla-ai.github.io/llamafile/troubleshooting/

Provenance

• Base executable: llamafile 0.9.3
• Model weights: unsloth/medgemma-1.5-4b-it:UD-Q4_K_XL

---

The use of this model is governed by the Health AI Developer Foundations terms of use.