Qdrant: Search

Run a vector search against a Qdrant collection. Supports plain dense search and hybrid retrieval (dense + sparse with RRF fusion). The sparse vector is computed on the fly from the request's query string using the configured sparse method (BM25 or SPLADE).

Pre-requisite: Install Qdrant application in Profile > {Organization} > Applications to grant Zparse access.

Parameters

Application—REQUIRED

Configured Qdrant application.

Collection—REQUIRED

Qdrant collection to search against. Must already exist.

Mode—REQUIRED

• Single Embedding — plain dense search. - Hybrid Search — dense + sparse prefetches fused with Reciprocal Rank Fusion (RRF). Pair with Sparse Method.

Dense Vector Name

Name of the dense vector when the collection uses named vectors. Defaults to dense.

Sparse Vector Name

Hybrid Search only — name of the sparse vector when the collection uses named vectors. Defaults to sparse.

Sparse Method

Hybrid Search only — algorithm used to derive the sparse query vector from the query text: - BM25 — classic lexical scoring. - SPLADE — learned sparse representation.

Dense Only

Hybrid Search only — when enabled, skip the sparse prefetch and run a named-dense search over Dense Vector Name. Useful when sparse retrieval would not help but the collection still uses named vectors.

Input

JSON—REQUIRED

Search request containing the dense query vector and (for hybrid) the original query text used to derive the sparse vector.

Output

JSON

Qdrant search response (matched points with scores and payloads).