Qdrant

Qdrant is an open-source vector search engine written in Rust, designed for AI retrieval applications such as retrieval-augmented generation, recommendation systems, semantic search, and AI agents. It provides vector similarity search through REST and gRPC APIs, with official clients including Python and JavaScript. The engine uses a custom storage engine called Gridstore and supports real-time indexing, so new vectors become searchable as soon as they are added.

Core capabilities include native hybrid search combining dense and sparse vectors (with BM25, SPLADE++, and miniCOIL), multivector support, JSON metadata filtering with nested, text, geo, and has_vector options, one-stage filtering during HNSW traversal, and reranking features such as score boosting, late interaction models like ColBERT, and Maximum Marginal Relevance. Quantization options include asymmetric, scalar, and binary approaches. A built-in web UI supports collection exploration and query testing, and Qdrant Cloud Inference can generate text and image embeddings.

Qdrant is available as open-source software, a fully managed Qdrant Cloud service on AWS, GCP, or Azure, Qdrant Hybrid Cloud for customer-managed Kubernetes, Private Cloud for air-gapped deployments, and Qdrant Edge in beta. Enterprise features include SSO, RBAC, private networking, audit logging, and compliance documentation. A free start option is available through Qdrant Cloud, with premium tiers covering additional security controls.