Structured outputs for Claude models available on Amazon Bedrock

Amazon Bedrock has introduced structured outputs for all Anthropic Claude models, available through both the bedrock-runtime and bedrock-mantle endpoints. This feature allows developers to receive responses that automatically conform to a predefined JSON schema, eliminating the need for manual parsing or post-processing. The capability is accessible whether using boto3, the bedrock-runtime APIs, or the Messages API with the bedrock-mantle endpoint.

The structured outputs feature enables reliable automation across industries by ensuring responses match a defined schema. Use cases include extracting required fields from financial earnings reports and compliance documents, parsing clinical notes into Electronic Health Records, and enriching product catalogs with validated data. Additional applications include analyzing contracts for key terms and obligations, as well as powering intelligent support ticket routing based on intents, sentiments, and entities.

Structured outputs are now supported where Claude models are available on Amazon Bedrock. Developers can implement this feature by referencing the Amazon Bedrock documentation for guidance on schema definition and API integration. The documentation provides instructions for configuring structured outputs in applications.

Availability of structured outputs varies by region, with support tied to the availability of Claude models. Developers should consult the regional availability details page to confirm support in their specific location before implementation. This ensures compatibility and access to the feature where required.