API Reference

Rank passages by their relation to a query.

post
https://ai.api.nvidia.com/v1/retrieval/nvidia/nemotron-3.5-rerank-8b/reranking

Order a list of passages by their relation to a query.

Each query + passage pair, including image tokens, must fit within the model's context window of 10240 tokens. By default (truncate: NONE) an over-length pair is rejected; set truncate: END to trim the end of the passage text until the pair fits.

Recent Requests
Log in to see full request history
TimeStatusUser Agent
Retrieving recent requests…
LoadingLoading…
Body Params
string
required
Defaults to nvidia/nemotron-3.5-rerank-8b

The model to use for ranking.

query
object
required

The query to rank the passages against.

passages
array of objects
required
length between 1 and 512

The list of passages to rank. Each passage is text, an image, or both.

Passages*
string
enum
Defaults to NONE

How to handle a query + passage pair that exceeds the 10240-token context window. NONE rejects the request. END removes tokens from the end of the passage text until the pair fits; the query and the image are always kept whole.

Allowed:
Responses

Updated about 1 hour ago

Did this page help you?
Language
Credentials
Bearer
LoadingLoading…
Response
Click Try It! to start a request and see the response here! Or choose an example:
application/json

Updated about 1 hour ago

Did this page help you?
country_code