rerank_stage

langchain_mongodb.pipelines.rerank_stage(query: str, path: str | List[str], num_docs_to_rerank: int, model: str | None = None) → List[Dict[str, Any]]

$rerank aggregation stage for Native Reranking in Atlas.

Requires MongoDB 8.3+ and Native Reranking enabled via Atlas Project Settings. Best used after a $search, $vectorSearch, $rankFusion, or $scoreFusion stage.

Will migrate to pymongo_search_utils once available there. (PYTHON-5876)

Parameters:

• query (str) – Text query used for reranking

• path (str | List[str]) – Field or list of fields to rerank on

• num_docs_to_rerank (int) – Number of documents to pass to the reranker (max 1000)

• model (str | None) – Voyage AI reranking model (e.g. “rerank-2.5”, “rerank-2”, “rerank-2.5-lite”). Omit to use the latest available model.

Returns:

$rerank followed by $set to update the score field

Return type:

List of pipeline stages