Google Gemma 4 Brings Advanced AI to Edge Devices

Google has released Gemma 4, its most advanced open-weights AI model family, built on the same architecture as Gemini 3. The lineup includes four models — Effective 2B, Effective 4B, a 26B Mixture of Experts, and a 31B Dense — designed to run complex reasoning tasks on low-power devices like smartphones and Raspberry Pi computers. Key features include native function calling, multimodal support for images, video, and audio, and context windows up to 256K tokens. All models are available under a permissive Apache 2.0 license via Google Cloud, Hugging Face, Kaggle, and Ollama.