Search
Reranking
Reranking is an advanced feature that improves the quality and relevance of search results by using specialized models to re-evaluate and reorder the initial results. This second-pass approach significantly enhances result accuracy at the cost of slightly increased latency.
Advanced Configuration
Configuration Options:
- Reranking Models: Use any supported reranking model from our inference endpoint. See Inference Models for available models and their capabilities. Defaults to
"mixedbread-ai/mxbai-rerank-large-v2"when not specified. - Top-k Constraint: The
top_kvalue must be smaller than or equal to the first-stage retrieval count (your maintop_kparameter). Defaults tonull, which returns all results from the first stage (respects your maintop_kparameter). - Metadata Inclusion: Defaults to
false. Set totrueto include all metadata, or provide a list of specific metadata keys to include.
Next Steps
Now that you understand reranking, explore these related topics:
- Search Basics: Core search concepts and parameters
- Metadata Filtering: Combine reranking with metadata filtering
- Data Models: Understand how scores and results work
- API Reference: Complete API details
Pro Tip: Start with reranking disabled to establish baseline performance, then enable it for queries where quality matters more than speed.
Metadata Filtering
Learn how to use metadata filtering syntax to narrow search results and find exactly what you're looking for in your Vector Stores.
File Search
Learn how to search for complete files in your Vector Store, finding relevant documents ranked by overall relevance.
Last updated: July 15, 2025