Qdrant
used in
0
stacks
Rust-based vector similarity search engine and database with payload filtering.
Open-source (Apache 2.0) vector database written in Rust, exposing REST and gRPC APIs for storing vectors with JSON payloads. Supports hybrid search combining dense vectors with sparse/BM25 retrieval, filtering applied during HNSW traversal, multivector support, and quantization to reduce memory use. Available self-hosted or via managed Qdrant Cloud on AWS, GCP, and Azure, plus hybrid and private cloud options.
Strengths
- Single-stage metadata filtering during HNSW traversal rather than pre/post-filtering
- Quantization options (scalar, product, binary) cut memory footprint substantially
- Native hybrid search with dense, sparse, and late-interaction reranking
- Rust implementation with REST/gRPC APIs and official Python/JS clients
Tradeoffs
- No built-in embedding modules historically; you manage vectorization yourself (cloud inference is newer)
- Distributed mode requires manual sharding and replication decisions compared to fully elastic systems
- Less natural fit than pgvector for teams wanting vectors inside an existing Postgres stack