Structured Outputs for Claude Models Arrive on Amazon Bedrock

Hi, it's Shii-chan!

AWS What's New aws.amazon.com

This might free you from parsing LLM output by hand!

What was announced?

AWS What's New announced that Amazon Bedrock now supports structured outputs for all Anthropic Claude models.

Two endpoints are supported:

• bedrock-runtime (through boto3 and the bedrock-runtime APIs)
• bedrock-mantle (through the Messages API)

On either one, you can receive schema-compliant JSON that conforms to a structure you define.

Why it matters

When you pass LLM output to downstream code, a broken format causes trouble. The announcement says you get schema-compliant results without custom parsing, retry logic, or post-processing.

What changes

The examples in the announcement:

• Extracting every required field from financial earnings reports and compliance documents
• Parsing clinical notes into Electronic Health Records (EHR)
• Enriching product catalogs with consistent, validated results
• Analyzing contracts for key terms, parties, and obligations
• Routing support tickets so intents, sentiments, and entities always match your application's data model

Dive Deep

The announcement is short and doesn't cover how to specify schemas or any parameters. See the Amazon Bedrock documentation for details.

Availability matches the Regions where Claude models are available.

Wrap-up

• Structured outputs work with all Claude models on Bedrock
• Both bedrock-runtime and bedrock-mantle are supported
• Less parsing, retry, and post-processing code
• Available wherever Claude models are available

This is for anyone who wants LLM output to flow reliably into their systems!