TL;DR: LangChain's OpenAI integration (langchain-openai==1.7.2) now correctly handles streaming output, preventing items from drifting into separate blocks.
Summary: LangChain released version 1.7.2 of its OpenAI integration package, langchain-openai. This update specifically addresses a bug where streaming responses from OpenAI models could experience output_index drift, causing stream items to be incorrectly split across multiple blocks.
Why it matters: This fix ensures more reliable and accurate processing of streaming outputs from OpenAI models within LangChain applications. Developers should update to prevent potential data integrity issues when working with real-time model responses.
Source: github_releases