# What is Cross-encoder reranking? Definition and meaning

A model reads the query and each candidate together and produces a more accurate relevance score for the shortlist.

## Definition

**Cross-encoder reranking**: A model reads the query and each candidate together and produces a more accurate relevance score for the shortlist.

In short: Re-scoring top results with a joint query and document model.

## When it is used

Raising precision of the final context passed to a language model.

## Learn more

The full guide to **[Cross-encoder reranking](/tools/cross-encoder-reranking)** explains how it works, when to use it and how it compares with the alternatives, with examples. The [glossary](/glossary) lists every term on the site from A to Z.

Previous term: [CrewAI](/glossary/crewai) · Next term: [Cross-validation](/glossary/cross-validation)
