Gemma 3n

Google has announced new AI models and tools focused on generative media creation and advanced on-device AI capabilities

Gemma 3n is a new, mobile-first AI model developed with Qualcomm, MediaTek, and Samsung, optimized for fast, private, and multimodal on-device performance

MatFormer training allows dynamic trade-offs between quality and performance, enabling submodels with different memory footprints for specific use cases

Gemma 3n supports privacy-first, offline-ready AI, ensuring user data stays on-device and works without internet connectivity

The model excels in multimodal understanding, handling text, audio, images, and video, including high-quality speech recognition and translation

Gemma 3n offers enhanced multilingual capabilities, especially in Japanese, German, Korean, Spanish, and French

Flow, an AI filmmaking tool, integrates Veo, Imagen, and Gemini models to help users create cinematic scenes and narratives using natural language