Archive — US — techcrunch
The latest Google Gemma AI model can run on phones
Google unveiled its Gemma 3n AI model at Google I/O 2025, designed for efficient operation on phones, laptops, and tablets. This model supports audio, text, images, and videos and can work offline, enhancing privacy and reducing costs. The release also includes MedGemma for health analysis and SignGemma for translating sign language, although its licensing has faced criticism.
Key points
- Google announced Gemma 3n at Google I/O 2025 for use on mobile devices and laptops.
- Gemma 3n can process audio, text, images, and videos while functioning offline.
- Offline AI models like Gemma 3n are preferred for privacy and cost-effectiveness.
- MedGemma is another model for health-related text and image analysis.
- SignGemma aims to assist the deaf and hard-of-hearing by translating sign language.
- Despite some criticism regarding licensing terms, Gemma models have been downloaded millions of times.
Why it matters
- The trend towards efficient, device-based AI models aligns with the increasing focus on data privacy in the tech industry.
- Similar movements in AI emphasize offline capabilities to lessen reliance on cloud computing and enhance user security.
- Models like SignGemma reflect ongoing efforts to improve accessibility for disabled individuals, an important aspect of tech innovation.
The introduction of Gemma 3n and its companions signifies a significant shift towards practical, privacy-focused AI, potentially impacting user experience and accessibility in the tech landscape.