Google announces Gemma 4 open AI models, switches to Apache 2.0 license
Summary
Ars Technica reports Google has released Gemma 4, its open-weight AI models, and switched licensing to Apache 2.0. The update emphasizes local/offline deployment with four model sizes, including 26B MoE and 31B Dense, plus mobile-optimized E2B and E4B variants, improved latency, and native function calling. The change aims to give developers more deployment flexibility and data control while retaining cloud options.