We use cookies. Cookies and browser storage keep you signed in and remember your preferences. With your permission we also measure how the site is used. You can any time. See our cookie policy.
mongodb/rerank-3-liteRuns on your balance at the prices above. New accounts start with free usage.
Settings
VoyageAI by MongoDB: rerank-3-lite is a reranking model from VoyageAI by MongoDB. It belongs to the rerank category, and its job is to take a set of candidate documents and reorder them by relevance to a query, rather than to generate new text. It is positioned as a drop-in upgrade to rerank-2.5-lite, which means teams already using that model can move to this one without changing how they integrate it.
The model accepts text as input and produces a reranking as output. It has a context length of 32,000 tokens and a maximum output of 28,800. Because its purpose is scoring and ordering rather than generation, the useful thing it returns is a relevance ordering over the passages you supply, and it is built to balance latency against quality. Across domain evaluations it improves on rerank-2.5-lite by 0.94% NDCG@10 on average, with a further improvement of 1.86%. That measured gain makes it a low-risk replacement where the earlier lite model is already in production.
Practical use cases follow directly from those capabilities. It fits retrieval augmented generation pipelines, where a first stage retriever returns a broad set of candidates and the reranker narrows and orders them before they reach a language model. It suits search and site search, where the ordering of results matters more than the raw match. It also fits question answering over document collections, internal knowledge bases and support content, and any workflow that needs passages sorted by how well they answer a specific query. The large context budget means long candidate passages can be included without aggressive truncation, and the emphasis on latency makes it reasonable for interactive applications rather than only offline batch jobs.
You call it through api14's OpenAI-compatible API using the model id "mongodb/rerank-3-lite". The supported parameters list is empty, so pass the model id with your input and rely on the defaults rather than expecting tuning options on the request. Because the endpoint follows the OpenAI-compatible shape, existing client libraries and code paths can be pointed at api14 and this model id with minimal change.
It is worth noting that this is a reranker, not a generator. If your task is to write answers, summarize or hold a conversation, a text generation model is the right choice, and this model works best as a stage in front of one. If you need to score or order text against a query, it is the appropriate fit.
import OpenAI from "openai";
const client = new OpenAI({
baseURL: "https://api14.com/api/v1",
apiKey: process.env.API14_API_KEY,
});
const result = await client.post("/rerank", { body: {
"model": "mongodb/rerank-3-lite",
"query": "What is the capital of France?",
"documents": [
"Paris is the capital of France.",
"Berlin is in Germany."
],
"top_n": 1
} });
console.log(result.results);
This community has no topics yet. Open the first one about VoyageAI by MongoDB: rerank-3-lite.