Introducing Gemma 4 models on Amazon Bedrock

Short summary

Google DeepMind's Gemma 4 family is now available on Amazon Bedrock with three instruction-tuned variants: Gemma 4 31B (dense architecture), 26B-A4B, and E2B (mixture-of-experts). All variants feature built-in reasoning, native function calling, and multimodal text-image input capabilities. These open-weight models, released under Apache 2.0 license, are optimized for intelligence-per-parameter efficiency and production deployment.

• •Three Gemma 4 variants available: 31B dense, 26B-A4B, and E2B mixture-of-experts architectures
• •Built-in reasoning, function calling, and multimodal input support across text and images
• •Open-source under Apache 2.0, optimized for intelligence-per-parameter efficiency

Generated with AI, which can make mistakes.