Skip to content
AI.info

tools

Mixedbread Rerank

Reranks an initial set of search results with specialized models to improve relevance for complex queries.

Mixedbread Rerank

Mixedbread Rerank adds a second ranking stage to Mixedbread Store searches. It can use the default listwise model or a selected supported model, limit the returned results, and include selected metadata.

It is used by developers building search, retrieval-augmented generation, and knowledge-base applications through the API or CLI. Reranking is billed separately: semantic search with reranking costs $7.50 per 1,000 queries, while Toast 1 searches with reranking cost $2.50 per 1,000 queries.

Features

  • Enables reranking with a boolean search option
  • Uses mxbai-rerank-v3.1-listwise as the default model
  • Supports listwise and pointwise reranking models
  • Sets a reranking top-k limit
  • Includes all or selected metadata in results
  • Works through the Mixedbread Stores API
  • Supports reranking from the Mixedbread CLI

Use cases

  • Improve relevance for complex knowledge-base searches
  • Rerank retrieved passages for retrieval-augmented generation
  • Prioritize recent documents or primary sources
  • Refine search results after semantic retrieval
  • Search internal stores from command-line workflows

Pros

    Cons

      Pricing

      Starting price
      Free
      Pricing checked
      2026-09-19

      Starter

      Free

      • $5 one-time credits
      • 3 workspace users
      • 10 stores
      • 100 requests / minute
      • Community Slack support

      Scale

      $20/ month

      • $20 of credits included every month
      • Unlimited workspace users
      • 10,000 stores
      • 1,200 queries/min, 360 ingestion/min
      • Automatic backups
      • Priority Slack support (same-day SLA)

      Enterprise

      Custom

      • Volume-based discounts
      • Unlimited workspace users
      • Unlimited stores
      • Custom rate limits & SLA
      • Automatic backups, point-in-time recovery
      • Dedicated support team
      Official website