Medgemma 4b It

🚀 MedGemma Model

MedGemma is a collection of models trained for medical text and image comprehension. It offers two variants, 4B and 27B, which can help developers accelerate the development of healthcare - based AI applications.

🚀 Quick Start

Prerequisites

First, install the Transformers library. Gemma 3 is supported starting from transformers 4.50.0.

$ pip install -U transformers

Run model with the pipeline API

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

pipe = pipeline(
    "image-text-to-text",
    model="google/medgemma-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).raw)

messages = [
    {
        "role": "system",
        "content": [{"type": "text", "text": "You are an expert radiologist."}]
    },
    {
        "role": "user",
        "content": [
            {"type": "text", "text": "Describe this X-ray"},
            {"type": "image", "image": image}
        ]
    }
]

output = pipe(text=messages, max_new_tokens=200)
print(output[0]["generated_text"][-1]["content"])

Run the model directly

# pip install accelerate
from transformers import AutoProcessor, AutoModelForImageTextToText
from PIL import Image
import requests
import torch

model_id = "google/medgemma-4b-it"

model = AutoModelForImageTextToText.from_pretrained(
    model_id,
    torch_dtype=torch.bfloat16,
    device_map="auto",
)
processor = AutoProcessor.from_pretrained(model_id)

# 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).raw)

messages = [
    {
        "role": "system",
        "content": [{"type": "text", "text": "You are an expert radiologist."}]
    },
    {
        "role": "user",
        "content": [
            {"type": "text", "text": "Describe this X-ray"},
            {"type": "image", "image": image}
        ]
    }
]

inputs = processor.apply_chat_template(
    messages, add_generation_prompt=True, tokenize=True,
    return_dict=True, return_tensors="pt"
).to(model.device, dtype=torch.bfloat16)

input_len = inputs["input_ids"].shape[-1]

with torch.inference_mode():
    generation = model.generate(**inputs, max_new_tokens=200, do_sample=False)
    generation = generation[0][input_len:]

decoded = processor.decode(generation, skip_special_tokens=True)
print(decoded)

✨ Features

• Multimodal Capability: The 4B version supports both text and vision modalities, while the 27B version focuses on text.
• High Performance: Outperforms the base Gemma 3 models across various multimodal and text - only health benchmarks.
• Long Context Support: Can handle a context length of at least 128K tokens.

📦 Installation

$ pip install -U transformers

💻 Usage Examples

Basic Usage

The above quick - start code snippets show the basic usage of running the model with the pipeline API and running the model directly.

Advanced Usage

For more advanced usage, such as fine - tuning the model, refer to the following Colab notebooks:

• [Quick start notebook in Colab](https://colab.research.google.com/github/google - health/medgemma/blob/main/notebooks/quick_start_with_hugging_face.ipynb)
• [Fine - tuning notebook in Colab](https://colab.research.google.com/github/google - health/medgemma/blob/main/notebooks/fine_tune_with_hugging_face.ipynb)

📚 Documentation

Model information

MedGemma is a collection of Gemma 3 variants. The 4B version utilizes a SigLIP image encoder pre - trained on medical data. It has both pre - trained and instruction - tuned versions. The 27B version is trained only on medical text and is optimized for inference - time computation.

Model architecture overview

The MedGemma model is based on Gemma 3 and uses the same decoder - only transformer architecture. For more details, refer to the Gemma 3 model card.

Technical specifications

Inputs and outputs

Input:

• Text string, such as a question or prompt
• Images, normalized to 896 x 896 resolution and encoded to 256 tokens each
• Total input length of 128K tokens

Output:

• Generated text in response to the input, such as an answer to a question, analysis of image content, or a summary of a document
• Total output length of 8192 tokens

Performance and validation

MedGemma was evaluated on various multimodal classification, report generation, visual question answering, and text - based tasks.

Imaging evaluations

*Based on [ref](https://www.nature.com/articles/s41591 - 020 - 0842 - 3), presented as a 4 - way MCQ per example for skin condition classification. **On balanced split, see ref. ***Based on multiple datasets, presented as 3 - 9 way MCQ per example for identification, grading, and subtype for breast, cervical, and prostate cancer.

Chest X - ray report generation

Text evaluations

Citation

@misc{medgemma - hf,
    author = {Google},
    title = {MedGemma Hugging Face},
    howpublished = {\url{https://huggingface.co/collections/google/medgemma - release - 680aade845f90bec6a3f60c4}},
    year = {2025},
    note = {Accessed: [Insert Date Accessed, e.g., 2025 - 05 - 20]}
}

🔧 Technical Details

• The model uses grouped - query attention (GQA) in its attention mechanism.
• It supports a long context length of at least 128K tokens.

📄 License

The use of MedGemma is governed by the [Health AI Developer Foundations terms of use](https://developers.google.com/health - ai - developer - foundations/terms).

To access MedGemma on Hugging Face, you're required to review and agree to [Health AI Developer Foundation's terms of use](https://developers.google.com/health - ai - developer - foundations/terms). To do this, please ensure you're logged in to Hugging Face and click below. Requests are processed immediately. [Acknowledge license](https://huggingface.co/collections/google/medgemma - release - 680aade845f90bec6a3f60c4)

Resources

• [Model on Google Cloud Model Garden](https://console.cloud.google.com/vertex - ai/publishers/google/model - garden/medgemma)
• [Model on Hugging Face](https://huggingface.co/collections/google/medgemma - release - 680aade845f90bec6a3f60c4)
• [GitHub repository](https://github.com/google - health/medgemma)
• [Quick start notebook](https://github.com/google - health/medgemma/blob/main/notebooks/quick_start_with_hugging_face.ipynb)
• [Fine - tuning notebook](https://github.com/google - health/medgemma/blob/main/notebooks/fine_tune_with_hugging_face.ipynb)
• Patient Education Demo
• [Contact](https://developers.google.com/health - ai - developer - foundations/medgemma/get - started.md#contact)