Structured outputs for Anthropic Claude models on Amazon Bedrock
Amazon Bedrock now supports structured outputs for all Anthropic Claude models via constrained decoding
“Structured outputs use constrained decoding to enforce your JSON schema during token generation, not after the fact.”
Amazon Bedrock has added structured outputs support for all Claude models on both its bedrock-runtime and bedrock-mantle endpoints, using constrained decoding to enforce JSON schemas at token generation time rather than post-processing. This eliminates parsing errors, removes the need for retry logic, and reduces output token counts for high-volume workloads. The feature supports batch inference, cross-region inference, and streaming, making it production-ready for agentic and data-extraction workflows.