Frontier AI In Your Hands: On-Device Intelligence With Gemma and Gemini Nano

Опубликовано 10 ноября 2025, 18:00

This talk explores how Google DeepMind is bringing AI from the lab to the real world—on device, private, and fast. Google DeepMind traces the path from AlphaFold to DolphinGemma, a specialized model decoding decades of dolphin audio entirely on a phone and introduce Gemma: an open model family powering a vibrant community across devices and data centers. Tris unveils a new mobile-first architecture underpinning Gemma 3n (open) and the latest Gemini Nano on Pixel 10, delivering over 900 tokens/sec prefill on the same hardware as the prior generation. He highlights the Gemma 3n Impact Challenge and inspiring projects like assistive vision for the blind. Finally, he shares why small, specialized models excel at translating intent to action with ultra-low latency, and how you can start today via Hugging Face, Kaggle, and Android’s ML Kit GenAI APIs.

Find the resources you need to develop using AMD products: amd.com/en/developer.html

***

© 2025 Advanced Micro Devices, Inc. All rights reserved. AMD, the AMD Arrow logo, EPYC, ROCm, and AMD Instinct and combinations thereof are trademarks of Advanced Micro Devices, Inc.