Model
Gemma 4
Google DeepMind's most capable open models, built from Gemini 3 research to maximize intelligence-per-parameter and run on local and edge hardware. Ships in four sizes (E2B, E4B, 26B MoE, 31B dense) under an Apache 2.0 license, with audio and vision support across 140 languages.
Model metadata
Modalities: text, image, audio