ClovisReranker
A specialized reranking model that reassesses and reorders vector-search results. It improves relevance by 20–40% in RAG pipelines by jointly analyzing the query and each document.
Capabilities
InputWhat you can send
Maximum volume50 documents
Accepted formatsText queryText documents
OutputWhat the model produces
Output formatsIndexScoreDocument
Result orderMost to least relevant
Recommended volume10 to 30 documents