30% OFF Pro PlanClaim

MedGemma

Advanced AI Models for Medical Text and Image Analysis

Powered by Dr7.ai Medical AI Platform

MedGemma is Google's family of open-weight medical AI models built on Gemma 3. Explore versions, benchmarks and downloads, or try the free MedGemma interactive demo below.

🧠
5
open MedGemma models
4B
Multimodal (image + text)
27B
Text-only or multimodal
Latest version: MedGemma 1.5 4B · Jan 2026

Try MedGemma Interactive Demo

Experience the power of Medical Gemma 4B IT model for advanced medical text and image analysis

Dr7.ai API

Use Dr7's own medical AI assistant through an OpenAI-compatible API. No GPU or deployment to manage.

See the API

Want More Powerful AI Medical Diagnosis Features?

Upgrade to Pro version to unlock advanced AI models, unlimited conversations, professional medical image analysis and more powerful features

10+
AI Model Options
24/7
24/7 Support
∞
Unlimited Conversations

What is MedGemma

MedGemma is a family of open AI models from Google, built on Gemma 3 and trained to understand medical text and medical images. Google released the first MedGemma models on 20 May 2025; the newest release is MedGemma 1.5 4B (13 January 2026).

Built on the powerful Gemma 3 architecture, MedGemma has been optimized for healthcare applications, providing developers with robust tools to create innovative medical solutions using MedGemma's advanced capabilities.

As part of the Health AI Developer Foundations, MedGemma aims to democratize access to advanced medical AI technology, enabling researchers and developers worldwide to build more effective healthcare applications with MedGemma. You may also see it written as Med Gemma, Med-Gemma, MedGamma, Medgema or Google MedGemma — these all refer to the same model family. MedGemma is not the same as Med-Gemini, a separate Google research model that has not been released publicly.

Latest version
MedGemma 1.5 4B
Jan 2026

Adds 3D CT and MRI volumes, whole-slide pathology, comparison with prior chest X-rays, anatomy localization and lab-report data extraction.

MedGemma 1.5 4B in detail

Checked October 2026: there is no MedGemma 2 and no MedGemma 1.5 27B — MedGemma 1.5 exists only as a 4B model.

MedGemma versions and models

Every official MedGemma checkpoint published by Google, plus the related MedSigLIP image encoder.

MedGemma 1.5 4BLatest
Parameters
4B
Input
Text + images, incl. 3D CT/MRI and whole-slide pathology
Released
Jan 2026
Hugging Face repo
google/medgemma-1.5-4b-it
License
HAI-DEF terms

Updated 4B model; only available as instruction-tuned 4B.

Details: MedGemma 1.5 4B
MedGemma 4B-it
Parameters
4B
Input
Text + images
Released
May 2025
Hugging Face repo
google/medgemma-4b-it
License
HAI-DEF terms

Instruction-tuned — the recommended starting point.

MedGemma 4B-pt
Parameters
4B
Input
Text + images
Released
May 2025
Hugging Face repo
google/medgemma-4b-pt
License
HAI-DEF terms

Pre-trained base model for deeper experimentation.

MedGemma 27B text-it
Parameters
27B
Input
Text only
Released
May 2025
Hugging Face repo
google/medgemma-27b-text-it
License
HAI-DEF terms

Trained only on medical text; highest text-benchmark scores.

MedGemma 27B multimodal
Parameters
27B
Input
Text + images + EHR data
Released
Jul 2025
Hugging Face repo
google/medgemma-27b-it
License
HAI-DEF terms

One 27B model for text, medical records and images.

MedSigLIPRelated
Parameters
400M + 400M
Input
Image and text embeddings (448×448)
Released
Jul 2025
Hugging Face repo
google/medsiglip-448
License
HAI-DEF terms

Not a chat model: the medical image encoder behind MedGemma, for classification and image search.

Details: MedSigLIP

Sources: MedGemma 1.5 model card · MedGemma 1 model card · MedGemma Technical Report (arXiv 2507.05201) · MedGemma 1.5 Technical Report (arXiv 2604.05081) · Nature Medicine (2026) · MedSigLIP model card

Features

What MedGemma can do, according to Google's model cards

MedGemma model variants

🖼️

Multimodal models (4B, 27B, 1.5 4B)

Read medical images and text together. They use a SigLIP image encoder pre-trained on de-identified medical data — chest X-rays, dermatology, ophthalmology and histopathology images.

📄

Text-only model (27B)

Trained exclusively on medical text and optimized for medical reasoning, so it scores slightly higher than the 27B multimodal model on text benchmarks.

Key capabilities

  • ✓Medical image classification (radiology, pathology, dermatology, ophthalmology)
  • ✓Medical image interpretation and draft report generation
  • ✓Medical question answering and clinical reasoning
  • ✓3D CT/MRI and whole-slide pathology interpretation (MedGemma 1.5)
  • ✓Extracting structured data from lab reports and EHR text (MedGemma 1.5)

MedGemma benchmark results

Accuracy (%) or macro F1 as reported in Google's official model cards. 27B results use test-time scaling.

BenchmarkMedGemma 4BMedGemma 1.5 4BMedGemma 27B textMedGemma 27B multimodal
MedQA (4 options)64.469.187.785.3
MedMCQA55.759.874.270.2
PubMedQA73.468.276.877.2
MMLU medical70.069.687.086.2
MIMIC-CXR (macro F1, top 5 conditions)88.989.5—90.0
EyePACS fundus64.976.8—75.3

Sources: MedGemma 1.5 model card · MedGemma 1 model card

Use cases for MedGemma

🏥

Healthcare app development

Build AI-based applications using MedGemma that examine medical images, generate reports, and triage patients with MedGemma's advanced capabilities.

🔬

Medical AI research

Accelerate research with open access to MedGemma advanced medical AI models through various platforms and cloud services supporting MedGemma.

👨‍⚕️

Clinical support prototypes

Prototype decision-support, summarization and record-review tools, then validate them before any clinical use.

How to Use MedGemma

Run MedGemma yourself with Hugging Face or Vertex AI, or try Dr7's medical AI chat online. Need the Dr7 API? It's right below the demo.

1

Medical AI chat by Dr7.ai

Ask Dr7's own medical AI assistant a medical question or upload an X-ray, CT slice or skin photo. Sign in to start.

Open the demo
2

Download from Hugging Face

Sign in to Hugging Face, open the model page and accept the Health AI Developer Foundations terms (access is granted immediately). Then load it with Transformers 4.50 or newer. Google says every MedGemma model can run on a single GPU; the 27B models need a large one unless you quantize them.

MedGemma on Hugging Face
3

Google Cloud Vertex AI

Deploy MedGemma as a managed endpoint from Vertex AI Model Garden in your own Google Cloud project — Google's recommended path for production scale. Billed by Google Cloud.

Vertex AI Model Garden

Quick start with Hugging Face Transformers (from Google's model card)

# pip install -U "transformers>=4.50" accelerate torch pillow requests

from transformers import pipeline
from PIL import Image
import requests
import torch

pipe = pipeline(
    "image-text-to-text",
    model="google/medgemma-4b-it",  # or "google/medgemma-1.5-4b-it"
    torch_dtype=torch.bfloat16,
    device="cuda",
)

# Image attribution: Stillwaterising, CC0, via Wikimedia Commons
image_url = "https://upload.wikimedia.org/wikipedia/commons/c/c8/Chest_Xray_PA_3-8-2010.png"
image = Image.open(requests.get(image_url, headers={"User-Agent": "example"}, stream=True, timeout=30).raw)

messages = [{"role": "user", "content": [
    {"type": "image", "image": image},
    {"type": "text", "text": "Describe this X-ray"},
]}]

output = pipe(text=messages, max_new_tokens=300)
print(output[0]["generated_text"][-1]["content"])

Adapting MedGemma to your task

Prompt engineering

Use few-shot examples and break tasks into smaller subtasks to improve results.

Fine-tuning

Train on your own medical data, starting from Google's fine-tuning notebook on GitHub.

Agentic orchestration

Combine MedGemma with tools such as web search, FHIR data generators or other AI models.

Before you build

Validation required

MedGemma is not clinical-grade out of the box. Google's model card states that developers must validate and adapt it for their specific use case before relying on it.

Terms of use

Use of MedGemma is governed by the Health AI Developer Foundations terms of use, which you accept before downloading the weights.

MedGemma vs Med-PaLM 2 vs LLaVA-Med

Which medical AI models can you actually download, what they read, and their published MedQA scores.

MedGemma 4BGoogle
Downloadable?
Yes (accept HAI-DEF terms)
Input
Text + images
MedQA
64.4%
MedGemma 1.5 4BGoogle
Downloadable?
Yes (accept HAI-DEF terms)
Input
Text + images (incl. 3D CT/MRI)
MedQA
69.1%
More: MedGemma 1.5 4B
MedGemma 27B textGoogle
Downloadable?
Yes (accept HAI-DEF terms)
Input
Text
MedQA
87.7%
MedGemma 27B multimodalGoogle
Downloadable?
Yes (accept HAI-DEF terms)
Input
Text + images
MedQA
85.3%
Med-PaLM 2Google
Downloadable?
No — weights never released
Input
Text
MedQA
86.5%
More: Med-PaLM 2
LLaVA-Med v1.5Microsoft
Downloadable?
Yes (research use only)
Input
Text + images
MedQA
Not reported
More: LLaVA-Med v1.5

Scores come from each model's own paper or model card under different settings (for example, Med-PaLM 2 used ensemble refinement prompting and MedGemma 27B used test-time scaling), so treat them as indicative rather than a head-to-head test. LLaVA-Med was evaluated on medical image question answering (VQA-RAD, SLAKE, PathVQA), not MedQA.

Sources: MedGemma 1.5 model card · MedGemma 1 model card · Med-PaLM 2 (arXiv 2305.09617) · LLaVA-Med (arXiv 2306.00890) · microsoft/llava-med-v1.5-mistral-7b

Have a photo of a visible symptom?

The models described on this page are research tools for clinical imaging. If what you have is an everyday photo — a rash, a mole, a nail, a wound — AI Doctor Scan reads it and returns a free urgency level and a plain-language summary in about 30 seconds. It does not name a diagnosis, and it needs no account.

Try AI Doctor Scan — free

FAQ

Common questions about MedGemma

What are the differences between MedGemma 4B and MedGemma 27B?

MedGemma 4B is a 4-billion-parameter multimodal model that reads medical images and text. MedGemma 27B comes in two 27-billion-parameter versions: a text-only model trained exclusively on medical text, and a multimodal model that also handles images and electronic health record data. The 27B models score higher on medical text benchmarks (MedQA 87.7% for 27B text-only vs 64.4% for 4B), while 4B is much cheaper to run. MedGemma 1.5 4B (January 2026) is an updated 4B model.

Is MedGemma ready for clinical use out of the box?

No. Google's model card says MedGemma is not intended to be used without appropriate validation, adaptation or meaningful modification for a specific use case, and its outputs are not intended to directly inform clinical diagnosis, patient management or treatment decisions.

How can I download or access MedGemma?

MedGemma weights are on Hugging Face (for example google/medgemma-4b-it). Sign in, accept the Health AI Developer Foundations terms on the model page — access is granted immediately — and load the model with Transformers. You can also deploy it from Google Cloud's Vertex AI Model Garden.

What types of medical images can MedGemma process?

MedGemma's image encoder was pre-trained on de-identified chest X-rays, dermatology images, ophthalmology (fundus) images and histopathology slides. MedGemma 1.5 4B adds 3D CT and MRI volumes, multiple patches from whole-slide pathology images, and comparison of a chest X-ray with prior images.

How can I improve MedGemma's performance on a specific task?

Google lists prompt engineering (for example few-shot examples), grounding, agentic orchestration with other tools, and fine-tuning on your own data. The official google-health/medgemma GitHub repository includes a fine-tuning notebook.

When was MedGemma released?

Google released MedGemma 4B and MedGemma 27B text-only on 20 May 2025, around Google I/O. MedGemma 27B multimodal and the MedSigLIP image encoder followed on 9 July 2025, and MedGemma 1.5 4B on 13 January 2026. The peer-reviewed MedGemma paper appeared in Nature Medicine on 6 October 2026.

How well does MedGemma score on medical benchmarks?

On MedQA (USMLE-style, 4 options), Google reports 64.4% for MedGemma 4B, 69.1% for MedGemma 1.5 4B, 87.7% for MedGemma 27B text-only and 85.3% for MedGemma 27B multimodal (27B results use test-time scaling). For comparison, the base Gemma 3 4B model scores 50.7%.

Are there resources for fine-tuning MedGemma?

Yes. Google's google-health/medgemma GitHub repository provides a quick-start notebook and a fine-tuning notebook for Hugging Face, plus further tutorial notebooks, for example for CT and whole-slide pathology with MedGemma 1.5.

What hardware do I need to run MedGemma?

Google says all MedGemma models can run on a single GPU, and that MedGemma 4B can even be adapted to run on mobile hardware. The 27B models need a large GPU — Google's quick-start notes point to Colab Enterprise-class GPUs to run them without quantization.

Is there a MedGemma 2 or a MedGemma 1.5 27B?

Not as of October 2026. The newest release is MedGemma 1.5, which Google offers only as a 4B multimodal instruction-tuned model. The 27B models are still the original MedGemma 1 versions from May and July 2025.

Does MedGemma work in languages other than English?

MedGemma is built on Gemma 3, which supports over 140 languages, so it can read and answer in many languages. However, Google notes that its safety evaluations used primarily English prompts and its medical benchmarks are mostly English, so you should validate quality in your own language (for example Japanese or Spanish) before relying on it.

Is MedGemma the same as Med-Gemini or Med-PaLM 2?

No. Med-Gemini (2024, 91.1% on MedQA) and Med-PaLM 2 (2023, 86.5% on MedQA) are Google research models whose weights were never released. MedGemma is Google's open model family that anyone can download and fine-tune.