Rerank

POST /v1/rerank takes a query and a list of candidate documents and returns them reordered by relevance, each with a score. Use it as the precision stage at the end of a retrieval pipeline: a fast first stage (vector search, BM25 or your existing store) pulls a top-k pool, then /v1/rerank sharpens the ordering of that pool before you show results or feed them to a generator.

How it works today

The current reranker is a bi-encoder. It embeds the query and each candidate independently and scores them by cosine similarity, returning the ranked list truncated to top_k. Because the candidates can be embedded ahead of time, this is cheap and fast, which makes it a practical refinement pass over a top-k pool.

A bi-encoder scores the query and the document separately, so it cannot model fine-grained interactions between specific query and document tokens. A cross-encoder, which reads the query and document together in a single pass, captures those interactions and is the stronger reranker. A cross-encoder reranker is on the roadmap. We would rather state that plainly than overclaim the current stage.

Request

/v1/rerank is a flinq-native endpoint, called with a plain HTTP client. model is required.

curl

curl https://api.flinq.ai/v1/rerank \
  -H "Authorization: Bearer flq_your_key_here" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "flinq-pilot-wombat",
    "query": "Stahlbetondecke d = 20 cm, C25/30, XC1",
    "documents": [
      "Ortbetondecke Stahlbeton, Dicke 20 cm, Beton C25/30, Expositionsklasse XC1",
      "Filigrandecke Stahlbeton, Elementdicke 5 cm, statisch mitwirkend",
      "Trapezblechdecke verzinkt, Stützweite 3,0 m"
    ],
    "top_k": 3
  }'

Python

import requests

r = requests.post(
    "https://api.flinq.ai/v1/rerank",
    headers={"Authorization": "Bearer flq_your_key_here"},
    json={
        "model": "flinq-pilot-wombat",
        "query": "Stahlbetondecke d = 20 cm, C25/30, XC1",
        "documents": [
            "Ortbetondecke Stahlbeton, Dicke 20 cm, Beton C25/30, Expositionsklasse XC1",
            "Filigrandecke Stahlbeton, Elementdicke 5 cm, statisch mitwirkend",
            "Trapezblechdecke verzinkt, Stützweite 3,0 m",
        ],
        "top_k": 3,
    },
)
ranked = r.json()["data"]  # sorted by relevance_score, index points into documents

JavaScript

const r = await fetch('https://api.flinq.ai/v1/rerank', {
  method: 'POST',
  headers: {
    Authorization: 'Bearer flq_your_key_here',
    'Content-Type': 'application/json',
  },
  body: JSON.stringify({
    model: 'flinq-pilot-wombat',
    query: 'Stahlbetondecke d = 20 cm, C25/30, XC1',
    documents: [
      'Ortbetondecke Stahlbeton, Dicke 20 cm, Beton C25/30, Expositionsklasse XC1',
      'Filigrandecke Stahlbeton, Elementdicke 5 cm, statisch mitwirkend',
      'Trapezblechdecke verzinkt, Stützweite 3,0 m',
    ],
    top_k: 3,
  }),
});
const { data: ranked } = await r.json();

Response

{
  "object": "rerank",
  "model": "flinq-pilot-wombat",
  "data": [
    {
      "index": 0,
      "relevance_score": 0.8421,
      "text": "Ortbetondecke Stahlbeton, Dicke 20 cm, Beton C25/30, Expositionsklasse XC1"
    },
    {
      "index": 1,
      "relevance_score": 0.6133,
      "text": "Filigrandecke Stahlbeton, Elementdicke 5 cm, statisch mitwirkend"
    },
    {
      "index": 2,
      "relevance_score": 0.2987,
      "text": "Trapezblechdecke verzinkt, Stützweite 3,0 m"
    }
  ],
  "usage": { "prompt_tokens": 71, "total_tokens": 71 }
}

Each hit keeps the original index into your documents array, a relevance_score (cosine similarity, higher is more relevant), and the text.

Parameters

Parameter	Type	Required	Default	Description
model	string	yes	—	Pilot model id. Unknown ids return 404.
query	string	yes	—	The query the pool is scored against. Over the model’s input cap it returns 422 rather than being silently cut.
documents	string[]	yes	—	Candidate pool from your first retrieval stage. Per-model cap: 2048 (otter) / 256 (wombat). Typical rerank pools are 20–200 documents.
top_k	integer	no	all	How many of the reranked hits to return. Omit it to get the full pool reordered.
dimensions	integer	no	1024	Matryoshka slice to score on (1024/768/512/256/128). Useful to match the slice your index retrieves on.

Headers. Authorization: Bearer flq_... (required); Idempotency-Key (optional) deduplicates the charge on retries.

Response fields.

Field	Type	Description
data[i].index	integer	Position of the hit in your documents array — map results back through this, not through text.
data[i].relevance_score	float	Cosine similarity of query and document (higher = more relevant). Comparable across requests of the same model; not comparable across models.
data[i].text	string	The document text, echoed back.
usage.total_tokens	integer	Billed tokens (below).

Billing. Rerank is billed on tokens(query) + Σ tokens(documents), at EUR 0.40 / 1M. The query is charged once per request, so the pool size drives the cost. Keep pools pre-filtered to a top-k shortlist rather than reranking a whole corpus.