Call Xinference rerank models with a Jina AI-compatible format.
Xinference Rerank uses a Jina AI-compatible format to rank candidate documents by relevance. Select an available Xinference rerank model; AGTCloud converts the response to the Jina AI-style structure.
Provider settings in Dify and similar clients
In clients that ask you to select a rerank provider type, choose Jina AI and enter an available Xinference rerank model ID.
POST/v1/rerank
Call a Xinference reranking model with the Jina AI-compatible format. AGTCloud normalizes the response to the Jina AI structure, so the request method, fields, and response shape match Jina Rerank.
Use a Xinference Rerank model visible to the current API key. Keep the query clear and specific, keep candidate documents semantically complete and within the model's token limit, process large collections in batches, and use top_n to limit the returned results.
The Xinference Rerank model ID visible to the current API Key.
Length1 <= length
query*string
Query text used for ranking the relevance of candidate documents. It is recommended to use clear and specific expressions to avoid queries that are too broad or vague.
Length1 <= length
documents*array<string>
Documents to rerank. Keep each document complete and semantically clear, and within the selected model's token limit. Process large collections in batches.
Items1 <= items
top_n?integer
Number of highest-relevance documents to return. When omitted, all results are returned. Set this value to reduce unnecessary result transfer and processing.