Google’s Gemma 4 Runs Offline on Any Device, & Is Completely Free to Use

Key Highlights – Google released Gemma 4 today, and the highlighting feature is not that it is the largest model in the family, rather the smallest. The E2B and E4B variants are built specifically for edge devices i.e. phones, Raspberry Pi, and other low-power hardware. They run offline, handle speech, vision, and code generation without a cloud connection, and are designed to operate with low latency on hardware that most AI models cannot run at all. This is a meaningful shift in what “open AI model” means in practice. Most open model releases target developers with access to capable GPUs. […]