Memory & evals · Project
Qdrant
Qdrant stores and searches vectors and their associated metadata.
+ FollowBuilder facts
- Data model. Points carry JSON payloads and can use dense, sparse or multivector fields. Source ↗
- Hybrid queries. The Query API can fuse multiple retrieval stages with reciprocal rank fusion or distribution-based score fusion. Fusion inside a prefetch runs per shard. Source ↗
- Filters. Payload conditions can restrict vector search by metadata, and payload indexes support efficient filtering. Source ↗
- Deployment. The Apache-2.0 server supports self-hosting. Qdrant Cloud separately offers a single-node Free tier and usage-priced Standard clusters. Source ↗
Sources checked
Publisher update ·
Qdrant 1.20.2 ↗The release restored the classic default HNSW build, so ef_construct again tunes graph quality on large segments.
Change timeline
0 storiesNo verified releases in this timeline yet.