Hey there, tech fans! Nuked here, ready to share some exciting news about the newest AI models from Google.

Google’s family of open AI models, known as Gemma, is expanding. At Google I/O 2025, they unveiled Gemma 3n, a lightweight AI designed for smooth performance on your phones, laptops, and tablets. It can handle all sorts of media including audio, text, images, and videos, all while being available in preview.

This development is a big step toward offline AI models that don’t need to connect to the cloud. They’re more budget-friendly and help keep your data private since all processing happens right on your device. According to Google, Gemma 3n can run on devices with less than 2GB of RAM, sharing the architecture of Gemini Nano, and offering impressive performance.

Aside from Gemma 3n, Google is launching MedGemma through its Health AI Developer Foundations program. It’s tailored for analyzing health-related text and images, making it useful for various health apps. Meanwhile, SignGemma is on the horizon—a model to translate sign language into spoken language, primarily for American Sign Language and English, aiming to support deaf and hard-of-hearing communities.

Despite its success, Gemma has faced criticism over its licensing terms, which some developers find risky for commercial use. Nonetheless, the models have been downloaded over 150 million times, showing strong community interest and trust.