Google released Gemma 4, its most intelligent family of open AI models. These models utilize the same technology as the flagship Gemini 3 series. The family includes four sizes, ranging from mobile-compatible versions to large-scale workstation models.
The models support advanced reasoning, code generation, and multi-step agentic workflows. Native multimodal capabilities allow the processing of images and video. Smaller versions also include audio inputs for on-device speech understanding.
Google released the entire Gemma 4 family under the Apache 2.0 license. This shift makes the models fully open-source for commercial and private use. Previous Gemma versions operated under more restrictive licensing terms.