Google releases Gemma 4, a family of open models built off of Gemini 3

Google has released Gemma 4, a family of open-weight models derived from Gemini 3.

The lineup includes 2B and 4B Effective models for edge devices and 26B Mixture of Experts and 31B Dense for larger machines.

All models handle images and video, the smaller ones also process audio, and Google says the larger variants ranked on Arena AI's text leaderboard.

Gemma 4 is released under the Apache 2.0 license and model weights are available via Hugging Face, Kaggle, and Ollama.