Google releases Gemma 4, a significantly improved multimodal open-source model that substantially outperforms its Gemma 3 predecessor across all benchmarks.

Gemma 4 represents a major advancement in small multimodal open models, delivering improvements across all performance metrics compared to Gemma 3

The model maintains Google's commitment to open-source AI development and accessibility

This update addresses limitations of previous versions and positions Gemma 4 as a competitive option in the small model space