Qdrant

Qdrant is an open-source vector database and vector search engine from Berlin. André Zayarni and Andrey Vasnetsov started the project in 2021, and Qdrant Solutions GmbH has since raised a Series B of 50 million US dollars. The engine is written in Rust and licensed under Apache 2.0. Developers use it to find similar items by meaning, for example in recommendation systems and retrieval-augmented generation.

Features

• Filtering on JSON payload that is attached to vectors, including keyword, full-text, numeric range and geo conditions
• Hybrid search with dense and sparse vectors and fusion methods such as reciprocal rank fusion
• Built-in quantization, which the project says can cut memory use by up to 97 percent
• Sharding and replication for horizontal scaling, self-hosted or as Qdrant Cloud with a free tier

Good to know

Qdrant is a specialised component for similarity search. For classic website search with typo tolerance and facets, a tool such as Meilisearch fits more directly. With about 35,000 GitHub stars it is one of the widely used open-source vector databases.