At the forefront of Google’s latest innovations showcased at the I/O 2025 event is the revolutionary Gemma 3n AI model. This cutting-edge technology is engineered to integrate seamlessly with smartphones, laptops, and tablets across various platforms. Gemma 3n stands out with its versatility, as it adeptly manages voice, text, images, and video data.
Highlighting its efficiency, Gus Martins, Gemma’s Product Manager, revealed that the model is capable of functioning even on devices with a mere 2 GB RAM. This is thanks to its shared architecture with the Gemini Nano model. The standout feature of Gemma 3n is its ability to operate entirely on device, eliminating the need for cloud processing. This not only enhances user privacy by keeping data local but also reduces high transaction costs associated with remote servers.
Despite its capability to run on mobile devices, Gemma 3n sacrifices nothing in terms of power. The model employs a groundbreaking technology from Google DeepMind known as Per-Layer Embeddings (PLE). This innovation minimizes memory requirements, making Gemma 3n 1.5 times more responsive on mobile devices compared to its predecessor, Gemma 3 4B.
Google’s event also spotlighted MedGemma, a specialized model tailored for the healthcare technology sector. MedGemma is lauded as the most advanced open model for interpreting health-related text and imagery. In parallel, Google is developing SignGemma, a model poised to transform the translation of sign language into spoken language. Trained on American Sign Language (ASL) and English, SignGemma is set to pave the way for next-generation applications benefiting the hearing-impaired community.
Yorum Yap