MOUNTAIN VIEW: Google has just released something huge—and it’s free. Meet MedGemma, a next-gen AI model powerful enough to analyse chest X-rays, medical images and patient histories with the accuracy of a trained radiologist. The kicker? It’s open-source and available for anyone to use.
The newly launched MedGemma 27B scored an impressive 87.7% on the MedQA benchmark—a gold standard in medical AI—and even the smaller version has been validated by US radiologists at 81% accuracy. That’s more than good enough to assist in real-world patient care.
But what sets this move apart is access. While most advanced medical AIs are hidden behind expensive paywalls or cloud platforms, Google made MedGemma and its sibling model MedSigLIP free to download, modify and run locally. That means hospitals, researchers and startups can now use cutting-edge diagnostic tools without compromising data privacy or blowing through budgets.
The models are built to understand both visuals and text. So MedGemma can not only scan an X-ray but also interpret patient history or a doctor’s note, making it a powerful digital second opinion for clinicians.
Google’s decision is already sending shockwaves through the healthcare AI world. MedSigLIP adds even more muscle by combining visual and written data to sharpen diagnostic accuracy further.
For regions with limited medical resources or tight budgets, this could be a game-changer. The potential to scale affordable, AI-driven diagnostics is now firmly within reach—and it’s open to the world.
