Reranker
A model that re-scores an initial retrieval set (from vector search or BM25) to prioritize the most relevant chunks for the current query. Cross-encoders (Cohere Rerank, BGE) are the standard implementation. Materially improves RAG groundedness.
Related terms
- Retrieval-Augmented Generation (RAG)
A pattern where an LLM answers a query by first retrieving relevant documents from an external corpus (typically via a vector database or hybrid search) and injecting them into the prompt as grounding context. RAG addresses freshness and factual grounding without model retraining.
- Hybrid Search
A retrieval strategy combining vector similarity search with sparse (BM25 or full-text) search, typically with a re-ranking step. In 2026 the reference RAG retrieval pattern — pure vector-similarity is a starter pattern.
Related on this site
Where this fits
Reranker is part of the AI Fundamentals vocabulary used in the Generative AI Maturity Framework. See the full glossary for the complete set of 149 defined terms, or take the free maturity assessment to see where your organisation stands.