Google DeepMind's Gemma 4 Models Now on Amazon Bedrock

Amazon Bedrock has expanded its offerings by integrating the Gemma 4 family of models, developed by Google DeepMind. These open-weight models are now accessible to developers and businesses through Bedrock, operating under the permissive Apache 2.0 license. The Gemma 4 series includes several instruction-tuned variants, such as Gemma 4 31B, 26B-A4B, and E2B, which utilize both dense and mixture-of-experts (MoE) architectures for efficient performance. Key features of these models include advanced reasoning capabilities, native function calling, and support for multimodal inputs encompassing both text and images, making them versatile for various AI applications.