# What changed in langchain-aws

> 15 changes in langchain-aws 1 from its official release notes, upgrade guides and advisories: 4 new APIs and options. Each one cited. Sources checked 2026-10-28. HTML: https://unstale.sh/pypi/langchain-aws/

To see which of these apply to a repository, run `npx @unstale/cli init` in it.

## langchain-aws 1

### New APIs and options

- Adds support for Z.ai models on ChatBedrockConverse. (1.9.1) — [release notes](https://github.com/langchain-ai/langchain-aws/releases/tag/langchain-aws%3D%3D1.9.1)
- Supports requestMetadata on InvokeModel API calls. (1.9.1) — [release notes](https://github.com/langchain-ai/langchain-aws/releases/tag/langchain-aws%3D%3D1.9.1)
- Model profiles declare mid-conversation support. (1.8.2) — [release notes](https://github.com/langchain-ai/langchain-aws/releases/tag/langchain-aws%3D%3D1.8.2)
- Claude Haiku 5.5 supports effort levels and default thinking. (1.8.2) — [release notes](https://github.com/langchain-ai/langchain-aws/releases/tag/langchain-aws%3D%3D1.8.2)

### Behavior and default changes

- Reports Converse thinking tokens in output_token_details. (1.9.0) — [release notes](https://github.com/langchain-ai/langchain-aws/releases/tag/langchain-aws%3D%3D1.9.0)

### Fixes and other changes

- Fixes the put→get round-trip for AgentCoreMemoryStore. (1.9.1) — [release notes](https://github.com/langchain-ai/langchain-aws/releases/tag/langchain-aws%3D%3D1.9.1)
- Replaces blanket media type profile overrides with per-model ones. (1.9.1) — [release notes](https://github.com/langchain-ai/langchain-aws/releases/tag/langchain-aws%3D%3D1.9.1)
- Adds missing reasoning_effort_levels overrides to model profiles. (1.9.1) — [release notes](https://github.com/langchain-ai/langchain-aws/releases/tag/langchain-aws%3D%3D1.9.1)
- Makes the fallback name for unnamed file blocks deterministic. (1.9.1) — [release notes](https://github.com/langchain-ai/langchain-aws/releases/tag/langchain-aws%3D%3D1.9.1)
- Prevents failure when content has a general plain-text MIME type. (1.9.0) — [release notes](https://github.com/langchain-ai/langchain-aws/releases/tag/langchain-aws%3D%3D1.9.0)
- Resolves actor_id for nested subgraph get_state. (1.9.0) — [release notes](https://github.com/langchain-ai/langchain-aws/releases/tag/langchain-aws%3D%3D1.9.0)
- Adds missing reasoning_effort_levels to model profiles. (1.9.0) — [release notes](https://github.com/langchain-ai/langchain-aws/releases/tag/langchain-aws%3D%3D1.9.0)
- ChatBedrock input_tokens now includes cache read and write tokens, matching ChatBedrockConverse behavior since 1.4.6. (1.9.0) — [release notes](https://github.com/langchain-ai/langchain-aws/releases/tag/langchain-aws%3D%3D1.9.0)
- Streaming support is fixed for Claude Haiku 5.5. (1.8.2) — [release notes](https://github.com/langchain-ai/langchain-aws/releases/tag/langchain-aws%3D%3D1.8.2)
- Streamed tool_use blocks with input that is not valid JSON are replayed. (1.8.2) — [release notes](https://github.com/langchain-ai/langchain-aws/releases/tag/langchain-aws%3D%3D1.8.2)