zerank-2

Each row is one provider serving these weights at one precision. Prices are per million tokens.

What BF16 means

What a reranker does

A reranker does not chat and does not generate text. You give it one question and a list of documents, and it scores how well each document answers that question — reading the pair together, rather than comparing two separate embeddings. That is why it catches relevance a vector search misses, and why it only makes sense on a shortlist: it is far more accurate per comparison, and far more expensive than a vector lookup.

Where it goes in a search pipeline

1. 1. Retrieve broadly. Your vector search or keyword index returns the top 50–100 candidates. Fast, cheap, approximate.
2. 2. Rerank. Send the query and those candidates here. You get them back ordered by true relevance.
3. 3. Use the top few. Put the best 3–5 in your prompt instead of the best 20. Better answers, a shorter prompt, and a smaller bill from whichever model writes the reply.

What people use it for

• RAG that cites the right page — the most common use. Retrieval finds twenty plausible chunks; the reranker picks the ones that actually contain the answer.
• Support and helpdesk search — matching a customer’s words to articles written in different words.
• Agent tool results — trimming a large search or file-read result down to what is worth spending context on.
• Site and catalogue search — reordering results before they reach the page.

Reading the scores

Every result carries a relevance_score between 0 and 1 and an index pointing back into the array you sent, so you can reorder your own objects without round-tripping the text. Results come back sorted, best first. Scores are calibrated to the model author’s own published scale, so thresholds you tuned elsewhere carry over unchanged.

Call it

Rerankers answer POST /v1/rerank — the Cohere and Jina shape, which is what the ecosystem settled on. There is no chat endpoint for this model, and sending it messages returns a 404.

curl https://api.infersia.com/v1/rerank \
  -H "Authorization: Bearer $INFERSIA_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "zeroentropy/zerank-2",
    "query": "how do I rotate an API key?",
    "documents": [
      "Create a replacement key in the console, then revoke the old one.",
      "The Great Wall of China is over 21,000 km long.",
      "API keys are shown once at creation time."
    ],
    "top_n": 3,
    "return_documents": true
  }'

top_n limits how many results come back; return_documents echoes the text alongside each score if you would rather not keep your own copy. Billing is on input tokens only — the query is counted once per document, because that is how the model reads it.

Pin a provider with zeroentropy/zerank-2@infersia, or take the cheapest with the bare id. Add :free for the rate-limited free tier where one exists.