jina-reranker-v3.5

AI model · Reranker / retrieval · Text → Tables / structured data by Jina AI

Domain-ready listwise reranker for multilingual and structured retrieval

jina-reranker-v3.5 is a domain-ready listwise reranker for multilingual and structured retrieval, featuring a 131K context window and 0.6B parameters for high-precision document ranking.

Model specs

The full spec card (context, pricing per token, modalities and more) on AntSpace. Sign up free to see the full spec card.

Launched
2026-07-27

Category: Machine Learning

Modalities
Text → Relevance Scores
Context window
131K tokens
Parameters
0.6B
Access
Jina API, Hugging Face, AWS, Azure, GCP, Elastic, Air-gapped
Released
2026-07-27
Parameter Count
0.6B
  • Domain-ready listwise reranker for multilingual and structured retrieval
  • 131K token context window for long-form content analysis
  • 0.6B active parameters for state-of-the-art ranking performance
  • Released on July 27, 2026

Overview

jina-reranker-v3.5 is a domain-ready listwise reranker model designed to improve the relevance of search results in multilingual and structured retrieval systems. It is intended for businesses that need to refine the output of initial search queries to ensure the most pertinent documents are ranked highest. The model supports an extensive 131K context window, making it capable of analyzing long-form content and complex data structures. With 0.6B parameters, it offers state-of-the-art ranking performance while remaining efficient for production use. jina-reranker-v3.5 is available through the Jina Reranker API and can be deployed on major cloud platforms including AWS, Azure, and Google Cloud.

Reranking Capabilities

  • Listwise Reranking: The model is a domain-ready listwise reranker designed for high-precision document ranking.
  • Multilingual & Structured Retrieval: It supports multilingual retrieval and is optimized for structured data search.
  • 131K Context Window: The model features a large context window of 131,000 tokens for processing long documents or large sets of candidates.
  • 0.6B Parameters: With 0.6B parameters, it provides a balance of high accuracy and fast inference speed.

Use cases

  • Businesses refine the output of initial search queries to ensure the most pertinent documents are ranked highest.
  • Developers process long-form content and complex data structures using the 131K token context window.
  • Organizations implement multilingual retrieval systems optimized for structured data search.

Specifications

Modalities
Text → Relevance Scores
Context window
131K tokens
Parameters
0.6B
Access
Jina API, Hugging Face, AWS, Azure, GCP, Elastic, Air-gapped
Released
2026-07-27
Parameter Count
0.6B
Context Length
131K tokens
Input Modality
Text (Query and Documents)
Output Modality
Relevance Scores

Added by AntSpace from public sources (jina.ai).

Related

More AI models by Jina AI