Combine full-text and vector search in Manticore using RRF to get more precise results than either method alone

Manticore

Hybrid search combines full-text (BM25) and vector (KNN) search in a single query, covering the blind spots of each method. Manticore Search implements this using Reciprocal Rank Fusion (RRF), which merges ranked result lists without needing to normalize scores across different scales. The post explains when hybrid search is useful (RAG pipelines, mixed structured/unstructured data, unpredictable query patterns), walks through the RRF formula with concrete examples, and shows how to use Manticore's `hybrid_match()` shorthand or explicit MATCH+KNN syntax. Tuning options include rank_constant, fusion_weights, and window_size. Multi-vector fusion is also supported, enabling simultaneous search across text, title embeddings, and image embeddings with per-signal weighting.

Hybrid search in Manticore Search

How hybrid search fits into modern search pipelines