Google Releases MedGemma, Built on Gemma 3, Specializing in Medical Text and Image Understanding
In the past two days, Google I/O 2025 has been very popular. In his keynote speech on the first day of the event, CEO Sundar Pichai shared many innovations, such as the full series upgrade of Gemini 2.5, the launch of Agent Mode on Chrome, the public beta of the coding agent Jules, the official release of Android XR, etc. Among the dazzling heavyweight updates,There is also a major open source achievement in the medical field hidden inside - MedGemma.
According to official introduction,The MedGemma series of models are based on Gemma 3.Contains Google's most powerful open source model for medical text and image understanding, available in 2 versions.There are four billion parameters for the multimodal version and 27 billion parameters for the plain text version.
The multimodal version of MedGemma 4B uses the SigLIP image encoder, which is specially pre-trained and uses data covering de-identified medical images, including chest X-rays, dermatology images, ophthalmology images, and histopathology sections. Its large language model component is trained based on a variety of medical data, including radiology images, histopathology image patches, ophthalmology and dermatology images, and medical text.and MedGemma 27B was trained specifically on medical texts and optimized for computations in the inference phase.
For medical text understanding and clinical reasoning scenarios, MedGemma can complete multiple tasks such as patient consultation, triage, clinical decision support, and medical text summarization. In order to facilitate readers to experience the powerful capabilities of MedGemma more intuitively,The "Tutorial" section of HyperAI's official website has now launched "One-click deployment of the MedGemma-27b-text-it medical reasoning model".Come and experience AI doctor consultation!
* Tutorial address:https://go.hyper.ai/Urygb
We have also prepared surprise benefits for new registered users. Use the invitation code "MedGemma" to register on the OpenBayes platform.You can get 4 hours of free use of RTX A6000 (the resource is valid for 1 month).Limited quantity, first come first served!
also,The tutorial section of the official website also launched "vLLM+Open WebUI Deployment II-Medical-8B Medical Reasoning Model".The model is based on Qwen3-8B and the model performance is optimized by using supervised fine-tuning using a medical domain-specific inference dataset and training DAPO (a possible optimization method) on a hard inference dataset.
* Tutorial address:https://go.hyper.ai/8fFFY
Demo Run
1. After entering the hyper.ai homepage, select the "Tutorial" page, select "One-click deployment of MedGemma-27b-text-it medical reasoning model", and click "Run this tutorial online".
2. After the page jumps, click "Clone" in the upper right corner to clone the tutorial into your own container.
3. Select "NVIDIA A6000 48GB" and "vLLM" image. The OpenBayes platform provides 4 billing methods. You can choose "pay as you go" or "daily/weekly/monthly" according to your needs. Click "Continue". New users can register using the invitation link below to get 4 hours of RTX 4090 + 5 hours of CPU free time!
HyperAI exclusive invitation link (copy and open in browser):
https://openbayes.com/console/signup?r=Ada0322_NR0n
4. Wait for resources to be allocated. The first clone will take about 2 minutes. When the status changes to "Running", click the jump arrow next to "API Address" to jump to the Demo page. Due to the large model, it will take about 3 minutes to display the WebUI interface, otherwise "Bad Gateway" will be displayed. Please note that users must complete real-name authentication before using the API address access function.
Effect Demonstration
The symptoms described by the author are: "I feel a little dizzy, want to vomit, and my throat is uncomfortable. What should I do?" It can be seen that MedGemma not only provides different solutions of "immediate medical treatment" and "self-care", but also provides the possible causes of this series of symptoms.
