Google released Gemma 4, calling it their "most intelligent open models to date". They are available in four sizes, Effective 2B (E2B), Effective 4B (E4B), 26B Mixture of Experts (MoE) and 31B Dense. Gemma 4 supports high-quality offline code and delivers frontier-level capability at far smaller sizes. E2B and E4B models are designed to run on phones.