---
title: "Jina AI's jina-reranker-v3 Ranks 64 Documents Together in One Pass"
subtitle: "Jina AI's 0.6B multilingual reranker beats models 10x its size on BEIR by scoring documents jointly in one context window."
url: "https://alphasignal.ai/news/jina-ai-s-jina-reranker-v3-ranks-64-documents-together-in-one-pass"
author: "AlphaSignal Newsroom"
author_url: "https://alphasignal.ai/editorial-team"
company: "Jina AI"
company_url: "https://alphasignal.ai/company/jinaai_"
published: "2026-09-27T10:00:19.810Z"
updated: "2026-09-27T10:03:39.340Z"
type: "Model"
topics:
  - "Retrieval"
  - "Benchmarks"
subtopics:
  - "Rag"
  - "Embeddings"
upvotes: 3731
source_url: "https://huggingface.co/jinaai/jina-reranker-v3"
comments: 0
truncated: true
---

# Jina AI's jina-reranker-v3 Ranks 64 Documents Together in One Pass

> Jina AI's 0.6B multilingual reranker beats models 10x its size on BEIR by scoring documents jointly in one context window.

Jina AI has released the [jina-reranker-v3 weights](https://huggingface.co/jinaai/jina-reranker-v3), a multilingual reranker with 0.6 billion parameters. The model encodes one query and as many as 64 candidate documents in a shared sequence, then assigns each candidate a relevance score. This listwise design lets later candidates incorporate information from earlier ones during a single model pass.

Second-stage reranking often requires running a cross-encoder separately for every query-document pair. Jina’s approach reduces that repeated computation and gives the model context about competing results. The small backbone also lowers weight memory, although long candidate lists can still consume substantial GPU memory.

## One sequence, many judgments

Jina calls the design “last but not late interaction” in its [architecture paper](https://arxiv.org/abs/2509.25085). A causal transformer processes the packed query and documents, while the final token of each document supplies its contextual representation. A lightweight multilayer projector maps that representation through dimensions of 1,024, 512, and 256 before producing the ranking output.

Causal attention makes the document order relevant. Later candidates can attend to the query and earlier documents, while earlier candidates receive no information from documents that follow them. Developers should test shuffled candidate orders because ranking stability may depend on how the first-stage retriever arranges results.

The 28-layer Qwen3-0.6B backbone supports a context window of roughly 131,000 tokens and up to 64 documents per call. Both limits apply, so long documents may exhaust the token budget before the list reaches 64 items. Attention cost also grows with the packed sequence length, which means the maximum context may exceed the practical memory or latency budget of a consumer GPU.

The optional 256-dimensional document embeddings are conditioned on the query and candidate list. They can support downstream analysis within that retrieval request, but they are unsuitable as static corpus embeddings for a vector index.

_Preview only. Full text requires an AlphaSignal Pro account._

## Sources

- [jinaai/jina-reranker-v3 on Hugging Face](https://huggingface.co/jinaai/jina-reranker-v3)
- [jina-reranker-v3: Last but Not Late Interaction for Listwise Document Reranking (arXiv 2509.25085)](https://arxiv.org/abs/2509.25085)
- [jina-reranker-v3 paper HTML](https://arxiv.org/html/2509.25085v1)
- [jinaai/jina-reranker-v3-GGUF](https://huggingface.co/jinaai/jina-reranker-v3-GGUF)
- [jinaai/jina-reranker-v3-mlx](https://huggingface.co/jinaai/jina-reranker-v3-mlx)
