langchain-ai/langchain-milvus

Enables multi-vector field storage and sparse embeddings within Milvus collections, supporting Milvus built-in functions like BM25 for keyword operations. Provides async-first APIs for retrieval operations and implements maximal marginal relevance filtering for diversity-aware result ranking. Integrates directly with LangChain's embedding and retriever abstractions, allowing drop-in vector store functionality across semantic search, recommendation, and RAG workflows.

52 stars and 556,475 monthly downloads. Used by 2 other packages. Available on PyPI.