meddoc

A vision + text medical assistant by Mittai17, built for rural and village settings in India. Talk to it in English, Tamil, or Tanglish and it replies in the same language.

Designed to give clear, safe, practical first-line advice and tell you exactly when to get professional help:

  • PRECAUTIONS โ€” what to do right now and what NOT to do
  • SOLUTIONS โ€” simple step-by-step instructions for home care (in order)
  • WHEN TO GET HELP โ€” warning signs meaning go to hospital or call an ambulance

Always includes the right emergency numbers: 112 (national), 108 (ambulance), 102 / 104 (health helplines), 100 (police), 101 (fire).

meddoc also understands images โ€” share a photo (skin, wound, X-ray, lab report) and it will describe what it sees and advise what to do next.

Files

File Description
medgemma1.5-meddoc-Q4_K_M.gguf GGUF model (Q4_K_M, 3.3 GB)
Modelfile Ollama Modelfile with the doctor system prompt

Use with Ollama

ollama create meddoc -f Modelfile
ollama run meddoc   # text
ollama run meddoc "what does this x-ray show?" (attach an image in Open WebUI)

Notes

Adapted from Google's MedGemma 1.5, which is subject to the Health AI Developer Foundations Terms of Use. meddoc is a starting point, not a clinical diagnosis tool โ€” when in doubt, go to a hospital.

Downloads last month
-
GGUF
Model size
4B params
Architecture
gemma3
Hardware compatibility
Log In to add your hardware

4-bit

Inference Providers NEW
This model isn't deployed by any Inference Provider. ๐Ÿ™‹ Ask for provider support