Sharper retrieval in one call

Rerank

Reorder candidate documents by relevance to a query with the Cohere-compatible /v1/rerank endpoint.

Get startedRead the docs
POST /v1/rerank
What is the capital of France?
  1. 1

    Paris is the capital of France.

    0.978
  2. 2

    France borders Spain and Italy.

    0.121
  3. 3

    Berlin is the capital of Germany.

    0.0001
01

Cross-encoder accuracy

Query and document read together.

02

Cohere compatible

Same request and response shape.

03

Sorted results

Highest relevance first.

04

Two-stage ready

Pairs with embeddings.

Built for

  • Better RAG

    Only the best passages reach the model

  • Search ranking

    Reorder keyword results

  • Answer selection

    Pick the best candidate

Rerank candidates
terminal
1curl https://api.llmgateway.io/v1/rerank \2  -H "Authorization: Bearer $LLM_GATEWAY_API_KEY" \3  -H "Content-Type: application/json" \4  -d '{5    "model": "qwen3-reranker-8b",6    "query": "What is the capital of France?",7    "documents": [8      "Paris is the capital of France.",9      "Berlin is the capital of Germany."10    ]11  }'
Read docs

More in models & modalities

All features

Moderations

Classify unsafe text and images

Response Healing

Malformed JSON, repaired

Image Generation

Every image model, one endpoint

Start using Rerank today

One API key. Every model. Live in minutes.

Start buildingBrowse models