Milvus v2

Milvus v2 is a cloud-native vector database built for trillion-scale similarity search in AI applications.

Milvus v2 redefines vector data management with a decoupled, cloud-native architecture that separates storage from computing. It handles massive unstructured datasets (think 10 billion vectors) by leveraging advanced indexing algorithms like HNSW and IVF_PQ. The platform supports hybrid search (combining scalar filtering with vector similarity) and guarantees high availability through its distributed design. Whether you are building a recommendation engine for 50 million users or a RAG pipeline using OpenAI embeddings, Milvus v2 delivers sub-millisecond latency at scale.

https://milvus.io/