LangChain-OpenAI 1.3.5 adds explicit prompt caching support

LangChain has released langchain-openai==1.3.5, a minor but impactful update to its popular OpenAI integration library. The headline feature is explicit prompt caching (via PR #38762), which allows developers to manually define cache keys for frequently used prompts. This moves beyond automatic caching to give fine-grained control over when and how responses are reused — critical for applications with heavy prompt repetition, like chatbots or RAG pipelines.

The release also includes a refresh of model profile data (PR #38774) to keep up with OpenAI's latest model offerings. The package, part of the larger LangChain ecosystem (141k stars on GitHub), streamlines building LLM-powered applications. This update is particularly useful for production deployments aiming to reduce API costs and latency without sacrificing flexibility.