langchain-anthropic==1.3.4
📦 langchainView on GitHub →
✨ 2 features🐛 5 fixes🔧 1 symbols
Summary
This release focuses on improving the Anthropic integration by adding Bedrock support and a User-Agent header, alongside several fixes related to model IDs and response handling.
Migration Steps
- If you were relying on the behavior related to OpenAI Responses block types, be aware that these are now filtered out.
- If you were using retired model IDs, update your configuration to use current model IDs.
✨ New Features
- Added ChatAnthropicBedrock wrapper.
- Added User-Agent header for Anthropic API calls.
🐛 Bug Fixes
- Filtered out common OpenAI Responses block types in Anthropic integration.
- Reverted addition of ChatAnthropicBedrockWrapper (Note: This was reverted, but a new feature addition suggests it might have been re-added or fixed later, based on the context of the surrounding fixes/features).
- Replaced retired model IDs in tests and docstrings.
- Set max input tokens based on 1m context beta header.
- Hoisted cache_control from tool_result content sub-blocks to tool_result level.