Full-text Search on Qdrant Cloud

Qdrant scores each result against your search terms using BM25, so keyword and semantic signals rank together. Start with a free cluster, no credit card.

TRUSTED BY ENTERPRISE TEAMS

Why Qdrant

Run Full Text BM25 Queries and Pass Candidates Into Hybrid Search

A text query on a payload field scored by BM25 over the full-text index, the top points passed as a prefetch into dense vector search, and the hybrid results ranked.
  • Run a BM25 query against any indexed payload field.

    Issue a text query on the payload field you have indexed for full text search and Qdrant scores results with BM25.

  • Pass BM25 candidates into dense vector search as a prefetch.

    A text query can act as a prefetch stage, letting you feed BM25 candidates directly into a dense vector search for full hybrid search in one pipeline.

  • Match points whose text contains at least one query term.

    A point matches the query when its text contains at least one of the terms you supply.

What you get

Prove Results Across Your Full Search Pipeline

Use full-text queries to rank results. Full-text queries contribute to how results are ranked, giving keyword signal a direct role in the final ordering.

Get a relevance score for every document that matches your search terms. For each document that matches the search terms, Qdrant calculates a relevance score based on how well that document matches.

Rank results directly from that relevance score. That score drives the final ordering of results, so keyword signal feeds directly into your ranked output.

Start free with Qdrant Cloud

Get started

Getting Started

Create a free cluster

in the Qdrant Cloud Console.

Set scoring to true on a full-text index to rank with BM25.

Create a full-text index on the payload field you want to search, then set scoring to true, and Qdrant ranks matching points using BM25.

Build the index from text already stored in your payload.

If your points already contain the field, Qdrant builds the index from the text stored in the payload, so you can index existing data in place.

Start Free

Common Questions

How do I get started with Qdrant Cloud?

A free cluster has one node with 1 GB of RAM and needs no credit card.

Will this work with hybrid search, combining keyword and semantic results in one query?

A text query can also be used as a prefetch, for example, to combine BM25 with dense vector search in hybrid search.

How does BM25 scoring work, and can we tune it?

Term frequency (TF): the more often a term appears in a document, the more relevant that document is likely to be.

If our text field is an array of objects, like reviews, does BM25 score each element separately or treat the whole array as one document?

When the elements are independent texts, such as reviews or passages of a long document, you may want to score each element as its own document instead.

Start Building With Full Text Search on Qdrant

Qdrant scores each result against your search terms using BM25, so keyword and semantic signals rank together.

Rocket flying over globe illustration