Amazon Bedrock adds structured JSON outputs for Anthropic Claude models
Amazon Bedrock now supports structured outputs for every Anthropic Claude model on both the bedrock‑runtime and bedrock‑mantle endpoints. When a request includes a JSON schema, the service validates the schema, compiles a grammar artifact, caches it for 24 hours, and then generates tokens that always produce a valid JSON document matching the schema. This eliminates the need for custom parsing,…
Key points
- Structured outputs now available for all Anthropic Claude models on Bedrock runtime and Mantle endpoints.
- JSON schema is enforced during generation, eliminating parse errors and reducing token usage.
- Compiled grammar artifacts are cached for 24 hours, so repeated schema calls have minimal overhead.
The feature works with batch inference, cross‑Region calls, and streaming of typed events. Because the model can only emit tokens that satisfy the schema, responses are deterministic, type‑safe, and often use fewer tokens, lowering inference costs for high‑volume workloads. Developers are advised to set "additionalProperties": false, use descriptive field names, and reuse schemas to benefit from caching. The first request for a new schema may be slower while the grammar is compiled, but subsequent calls are fast.
Model page: Claude Haiku 5.5 →
Structured outputs for Anthropic Claude models on Amazon Bedrock
AWS Machine Learning Blog · 19 October 2026
Loading the full article…
This text was published by AWS Machine Learning Blog and written by Anshika Saxena. It is reproduced here with attribution so you can read it in full; the rights remain with the publisher. Read it at the source ↗
The headline, key points and digest above were generated by Digest AI's editorial model from the linked sources. Automated summaries can contain errors: the sources are the record. Spotted a mistake? Tell us. Published by Martin K., who runs Digest AI and handles corrections.
More in Agents & Tools
All →- Amazon SageMaker Studio available via new AWS sign‑in experience for developers · 1 src
- Microsoft introduces Guardian Agents to monitor AI agents as insider threats · 4 src
- Amazon adds over 100 new models from 30 providers to SageMaker JumpStart in 2026 · 1 src
- Atlassian launches Agentic Multiplayer Protocol to tag AI agents with user attribution · 1 src
- Zillow adds Claude connector for listings, tours and financing · 1 src
Comments
via GitHub Discussions