langchain-aws==1.0.0a1
What's Changed
- Support multiple system messages for Anthropic by @michaelnchin in #572
- fix: Set bedrock_client region name to runtime client value when needed by @michaelnchin in #592
- chore: add
SECURITY.mdby @mdrxy in #636 - chore: add
CONTRIBUTING.mdby @mdrxy in #637 - fix(aws): (bedrock) support thinking blocks in streaming, multi-turn contexts by @ccurme in #638
- Add chat model integration tests for multiple system messages by @michaelnchin in #634
- fix: Optimize base model identification for application inference profiles by @michaelnchin in #615
- fix: Final assistant content cannot end with trailing whitespace by @michaelnchin in #599
- fix(bedrock): disallow forced tool_choice when thinking is enabled on Claude 3.7 by @jitokim in #640
- Allow all claude models with thinking enabled to accept tool_choice = "auto" + update associated code and tests by @awg66 in #646
New Contributors
Full Changelog: langchain-aws==0.2.33...langchain-aws==1.0.0a1