Milvus is an open-source system for storing and searching high-dimensional embeddings, designed to power retrieval-augmented and similarity-search applications. It offers quick local installs (pip) and client libraries to get prototypes running in minutes.

Multiple deployment modes fit different stages: a lightweight library for notebooks, a robust single-node option for testing, a horizontally scalable distributed cluster for production, and a fully managed Zilliz Cloud service for hassle-free operations. The platform is tuned to handle massive datasets while keeping search latency low.

Built-in features include metadata filtering, hybrid search, multi-vector queries, and integration with common AI toolchains for RAG, image and multimodal retrieval, and recommendation systems. Milvus is supported by extensive docs, community resources, and open-source tooling to accelerate development and productionization.