AWS announced the availability of Google DeepMind's Gemma 4 model family on Amazon Bedrock. Featuring dense and mixture-of-experts architectures, these models support complex workflows such as reasoning, multimodal understanding, agentic task execution, and software engineering, and process inputs across text, image, video, and audio modalities.
The three variants—Gemma 4 31B, 26B-A4B, and E2B—offer varying trade-offs between performance, latency, and cost to serve diverse use cases ranging from intensive coding tasks to interactive applications. They operate on a platform optimized for price-performance and enhanced capabilities like native function calling and streaming responses.
Source & further reading
Gemma 4 models now available on Amazon Bedrock ↗Source published 2026-06-10. This report was prepared with AI assistance from the linked primary-source material. Claims and limitations should be checked against the original announcement or research.