Google launched a new series of AI models in its Gemma collection at Google I/O. Gemma models are “open” in the sense that developers can build apps on top of them and fine-tune them, including offline.
Gemma 3n is a general-purpose model that can handle audio, text, images, and videos, and run on devices with as little as 2GB of RAM. MedGemma can analyze health-related text and images. As for SignGemma, it can translate sign language to spoken-language text; Google says it’s best at American Sign Language and English.
