[Release] langchain-ai/langchain langchain-anthropic==1.5.5: langchain-anthropic==1.5.5

LangChain released langchain-anthropic version 1.5.5, a patch update that fixes critical issues with reasoning token reporting in usage metadata, resolves a KeyError in Claude's file-tool middleware during rename operations, improves error messaging for missing credentials, and refreshes model profile data. This release ensures better observability of token usage and more robust file handling in Anthropic integrations.

Key Points

• •Reasoning tokens are now properly reported in usage metadata for better cost tracking and monitoring
• •Fixed KeyError that occurred when renaming files in Claude file-tool middleware
• •Improved error messaging to provide clearer guidance when Anthropic credentials are not configured
• •Model profile data has been refreshed to reflect latest Anthropic model capabilities
• •This is a patch release (1.5.4 → 1.5.5) focused on bug fixes and stability improvements
• •File handling in Claude integration is now more robust and error-resistant

Workflow Diagram

Concepts