About Qdrant
Qdrant is an open-source vector search engine written in Rust. It stores embeddings and returns the most similar items for AI retrieval, with filters on metadata.
What it does
Developers store vectors together with JSON metadata and query them through an API. The engine supports dense and sparse vectors, hybrid search that blends keywords and meaning, multivector records and reranking. It can run self-hosted or as a managed cloud service, so teams can match the setup to their needs.
Key features
- Dense and sparse vector search
- Hybrid search that blends keywords and vectors
- Metadata filters on nested, text and geo fields
- Multivector records for richer matching
- Self-hosted or managed cloud deployment
Who it's for
- Teams building retrieval for AI apps
- Engineers who need fast filtered search
- Companies with strict compliance needs
Pricing and good to know
The engine is open source under the Apache-2.0 licence, as listed on GitHub. A managed cloud option can be started for free. The cloud service is described as SOC 2 and HIPAA compliant. Self-hosting means you manage upgrades, backups and capacity yourself.
Highlights
- Search dense and sparse vectors
- Blend keyword and vector search
- Filter on nested, text and geo fields
- Store multivector records
- Run self-hosted or in the cloud
