Vector queries #

kNN similarity search over dense and sparse vectors, multi-vector fusion and hybrid retrieval.

• Vector query (kNN) — the k nearest neighbors of a query vector; text-in via server-side embedding, semantic_text wiring included
• Multi-vector query — several vector queries over different fields, fused into one ranking
• Semantic query — plain text in, the vector field and model resolve from the schema

SQL twins: KNN(field, […], k) covers the plain vector query (see the vector page); multi-vector fusion and schema-resolved semantic search stay DSL-side.