Google DeepMind has announced MedGemma, a collection of open models it describes as its most capable for health AI development. The new models are multimodal, meaning they can work with multiple types of input.
The release is aimed at developers in the medical and health space, offering openly available tools for building AI-powered applications. DeepMind did not specify concrete capabilities or benchmarks in the announcement.
This is a single-source update, so there is no independent comparison to prior models. The significance lies in expanding open access to health-focused AI models from one of the field's major labs.