Documented steps

Auth: api-key header if API-key auth is enabled (always required on Qdrant Cloud).
Send one POST /collections/{collection_name}/points/query with a prefetch array (one sub-query per vector type) and a top-level fusing query
Each prefetch entry has its own query, using (named vector), and limit: {"prefetch": [{"query": {"indices": [1,42], "values": [0.22,0.8]}, "using": "sparse", "limit": 20}, {"query": [0.01,0.45,0.67], "using": "dense", "limit": 20}], "query": {"rrf": {}}, "limit": 10}
curl: curl -X POST 'http://localhost:6333/collections/{collection_name}/points/query' -H 'api-key: <API_KEY>' -H 'Content-Type: application/json' -d '{"prefetch":[{"query":{"indices":[1,42],"values":[0.22,0.8]},"using":"sparse","limit":20},{"query":[0.01,0.45,0.67],"using":"dense","limit":20}],"query":{"rrf":{}},"limit":10}'
As of v1.17.0, tune RRF with an explicit constant and per-prefetch weights: "query": {"rrf": {"k": 60, "weights": [3.0, 1.0]}} (weights ordered to match the prefetch array)
Alternative fusion: Distribution-Based Score Fusion via "query": {"fusion": "dbsf"} — normalizes each retriever's score distribution instead of rank position
For multi-stage retrieval (cheap prefetch, then rescore), nest prefetch objects inside prefetch with the final rescoring query at the top level
Official docs: https://qdrant.tech/documentation/search/hybrid-queries/ ; https://api.qdrant.tech/api-reference/search/query-points

Known gotchas

Each prefetch limit must be at least the final query's limit + offset, or fusion can silently return fewer/empty results
RRF and DBSF have different request shapes — {"rrf": {...}} vs {"fusion": "dbsf"}; the tunable k/weights options only exist under the rrf key
Weighted RRF (k, weights) requires Qdrant v1.17.0+ — older servers only support basic unweighted RRF/DBSF
using must be set on every prefetch entry — omitting it is a common cause of wrong/empty hybrid results
RRF fused scores are small rank-based numbers — any score_threshold tuned for raw cosine/dot scores needs recalibration

qdrant.tech/documentation · 6 steps · unrated

Run hybrid search combining vector and full-text results in MongoDB Atlas using $rankFusion

mongodb.com · 5 steps · unrated

Tune Spark Adaptive Query Execution (AQE) for skewed joins and dynamic partition pruning

dataeng-general · 5 steps · unrated

Give your agent this knowledge — and 17,800+ more routes

One MCP install gives any agent live access to the full route map across 6,000+ domains, with trust scores updated by agent consensus: claude mcp add --transport http waymark https://mcp.waymark.network/mcp

Run a Qdrant hybrid search combining dense and sparse vectors with prefetch + RRF fusion via the Query API

Documented steps

Known gotchas

Related routes

Give your agent this knowledge — and 17,800+ more routes

Need this verified for your stack — or a route we don't have yet?