"
Overview
Qdrant is an open-source vector search engine written in Rust, purpose-built for high-performance, production-grade AI retrieval. It provides a fast and scalable vector similarity search service with a convenient API, enabling developers to ship full-feature vector search at any scale and with any deployment model. With over 30k GitHub stars and a community of 60k+ members, Qdrant has become a go-to infrastructure for teams building AI-powered applications that demand speed, accuracy, and reliability.
Developers love Qdrant for its production-ready architecture, which is engineered for real-time retrieval with the speed, accuracy, and scale that modern AI demands. Unlike traditional keyword-based search, Qdrant is designed for the AI era, supporting dense and sparse vectors, multivector models, and advanced filtering that is applied during HNSW traversal for high recall with low latency. Its open-source DNA is complemented by enterprise-grade security and flexibility, allowing teams to run on-prem, hybrid, edge, or seamlessly move to Qdrant Cloud.
Key Features
- Native Hybrid Search: Blend keyword and vector search in a single query using dense or sparse vectors, with support for BM25, SPLADE++, and miniCOIL.
- Expansive Metadata Filters: Store metadata in JSON and use advanced filters such as nested, text, geo, and has_vector for precise retrieval.
- Built-in Multivector: Make the retrieval layer more expressive, flexible, and multimodal by supporting multiple vectors per object.
- Efficient One-Stage Filtering: Filters are applied during HNSW traversal, eliminating pre- or post-filtering and ensuring high recall with low latency even under complex conditions.
- Full-Spectrum Reranking: Infuse business logic with score boosting, achieve token-level precision with late interaction models like ColBERT, and diversify results with Maximum Marginal Relevance (MMR).
- Flexible Deployment: Deploy anywhere at enterprise scale with options for Qdrant Cloud (fully,
