Qdrant is an open-source vector database and similarity search engine written in Rust. It stores vectors alongside JSON metadata and answers nearest-neighbor queries so applications can retrieve relevant items by meaning rather than exact keywords. The engine includes a custom storage layer (Gridstore), real-time indexing without full rebuilds, and quantization techniques that reduce memory use dramatically for large collections.
For retrieval quality, Qdrant supports native hybrid search that blends dense and sparse vectors in a single query — with BM25, SPLADE++ and miniCOIL — plus multivector storage, one-stage filtering during HNSW traversal, and reranking options such as score boosting, ColBERT-style late interaction and MMR diversification. Common applications include RAG and GenAI pipelines, AI agents with persistent memory, semantic search, recommendation systems, and anomaly detection.
Deployment is flexible: run the open-source engine yourself, or use Qdrant Cloud (managed, usage-based, on AWS, GCP or Azure), Hybrid Cloud (managed clusters in your own Kubernetes), Private Cloud (air-gapped and compliant setups) or the newer Qdrant Edge for low-latency local search. Free-tier cloud clusters, SOC 2 and HIPAA aligned posture, and enterprise tooling like SSO, RBAC and Prometheus/Grafana integrations make it suitable for both prototypes and production.
