Qdrant

Qdrant builds an open-source vector database that developers use for semantic search, recommendations, retrieval-augmented generation and AI agent memory. Qdrant Vector Database combines dense and sparse vector search, metadata filters, multiple vectors per object and reranking. Developers can combine these capabilities at query time to retrieve context for different workloads. Deployment options include managed Qdrant Cloud, Hybrid Cloud on customers’ Kubernetes infrastructure and Private Cloud for air-gapped environments. Cloud Inference generates text and image embeddings within the managed service.

Founded in 2021 by CEO André Zayarni and CTO Andrey Vasnetsov, Qdrant grew from work on matching unstructured data. Its engine is written in Rust. Vasnetsov’s earlier research on filterable HNSW explored adding graph connections within categories so similarity searches could enforce metadata constraints during traversal. That approach underpins Qdrant’s filtering architecture, applying conditions during search rather than as a separate pre- or post-filtering step.

In 2026, Qdrant announced a $50 million Series B led by AVP and reported more than 250 million downloads across its packages, with production users including Canva, HubSpot and Bosch. As of August 2026, Bayer’s myGenAssist ran on a four-node Qdrant Hybrid Cloud cluster holding roughly 135 million points across seven collections.