
Health AI Developer Foundations (HAI-DEF)
Groups models released for use in health AI by Google. Read more about HAI-DEF at https://developers.google.com/health-ai-developer-foundations
- Image-Text-to-Text • 29B • Updated • 21.6k • 183
google/medgemma-27b-text-it
Text Generation • 27B • Updated • 39.2k • 331google/medgemma-4b-pt
Image-Text-to-Text • 4B • Updated • 9.08k • 110google/medgemma-4b-it
Image-Text-to-Text • 5B • Updated • 111k • 626
google/medsiglip-448
Zero-Shot Image Classification • 0.9B • Updated • 16.3k • 63Note MedSigLIP is a SigLIP variant that is trained to encode medical images and text into a common embedding space. It was trained on a variety of de-identified medical image and text pairs, including chest X-rays, dermatology images, ophthalmology images, histopathology slides, and slices of CT and MRI volumes, along with associated descriptions or reports.
google/txgemma-9b-predict
Text Generation • 9B • Updated • 12.6k • 23google/txgemma-9b-chat
Text Generation • 9B • Updated • 3.31k • 39google/txgemma-27b-chat
Text Generation • 27B • Updated • 824 • 54google/txgemma-27b-predict
Text Generation • 27B • Updated • 29.9k • 34google/txgemma-2b-predict
Text Generation • 3B • Updated • 2.92k • 40
google/hear-pytorch
Image Feature Extraction • Updated • 344 • 10Note Health Acoustic Representations accelerates AI development for bioacoustic data e.g., coughs or breath sounds. The model is pre-trained on 300 million 2-second audio clips to produce embeddings that capture dense features relevant for bioacoustic applications.
google/hear
Updated • 60 • 27Note Health Acoustic Representations accelerates AI development for bioacoustic data e.g., coughs or breath sounds. The model is pre-trained on 300 million 2-second audio clips to produce embeddings that capture dense features relevant for bioacoustic applications.
google/path-foundation
Image Classification • Updated • 85 • 54Note Path Foundation accelerates AI development for histopathology image analysis. The model uses self-supervised learning on large amounts of digital pathology data to produce embeddings that capture dense features relevant for histopathology applications.
google/derm-foundation
Image Classification • Updated • 416 • 60Note Derm Foundation accelerates AI development for skin image analysis. The model is pre-trained on large amounts of labeled skin images to produce embeddings that capture dense features relevant for dermatology applications.
google/cxr-foundation
Image Classification • Updated • 262 • 84Note CXR Foundation accelerates AI development for chest X-ray image analysis. The model is pre-trained on large amounts of chest X-rays paired with radiology reports. It produces language-aligned embeddings that capture dense features relevant for chest X-ray applications.