other
0.278B parameters
Restricted
Reviewed May 2026

Jina Reranker v2 Base Multilingual

Jina Reranker v2 Base Multilingual is a 278M-parameter cross-encoder from Jina AI with a 1024-token context, trained on 100+ languages plus code and structured data (function-calling JSON, SQL). It is roughly 6x faster than bge-reranker-v2-m3 at comparable nDCG@10 on multilingual MIRACL but is gated under CC-BY-NC-4.0 for non-commercial use only.

License: cc-by-nc-4.0·Context: 1,024 tokens
BLK · VERDICT

Our verdict

OP · Fredoline Eruo|VERIFIED MAY 29, 2026
unrated

The fastest credible multilingual reranker, with the unusual addition of code and tool-call training data. The CC-BY-NC license eliminates it from any commercial product — use bge-reranker-v2-m3 instead unless you're prototyping or paying Jina for their API.

Overview

Jina Reranker v2 Base Multilingual is a 278M-parameter cross-encoder from Jina AI with a 1024-token context, trained on 100+ languages plus code and structured data (function-calling JSON, SQL). It is roughly 6x faster than bge-reranker-v2-m3 at comparable nDCG@10 on multilingual MIRACL but is gated under CC-BY-NC-4.0 for non-commercial use only.

Strengths

  • 278M params — ~6x throughput vs. bge-reranker-v2-m3 at similar multilingual quality
  • Trained on code retrieval, function-calling, and tabular data — not just prose
  • Multilingual 100+ languages with strong MIRACL nDCG@10 (~68)
  • Flash Attention 2 enabled, ONNX export shipped in repo

Weaknesses

  • CC-BY-NC-4.0 license — commercial deployment requires Jina API contract
  • 1024-token context vs. bge-v2-m3's 8192 — long passages need pre-truncation
  • Custom code path (trust_remote_code=True) required for the cross-encoder head
  • No fully open training data documentation — license risk for downstream fine-tuning

Quantization variants

Each quantization trades model quality for file size and VRAM. Q4_K_M is the most popular starting point.

QuantizationFile sizeVRAM required
Q4_K_M0.2 GB1 GB

Get the model

HuggingFace

Original weights

huggingface.co/jinaai/jina-reranker-v2-base-multilingual

Source repository — direct quantization required.

Hardware that runs this

Cards with enough VRAM for at least one quantization of Jina Reranker v2 Base Multilingual.

Compare alternatives

Models worth comparing

Same parameter band, plus what's one tier above and below — so you can decide what actually fits your hardware.

Step down
Smaller — faster, runs on weaker hardware
No verdicted models in the next tier down yet.

Frequently asked

What's the minimum VRAM to run Jina Reranker v2 Base Multilingual?

1GB of VRAM is enough to run Jina Reranker v2 Base Multilingual at the Q4_K_M quantization (file size 0.2 GB). Higher-quality quantizations need more.

Can I use Jina Reranker v2 Base Multilingual commercially?

Jina Reranker v2 Base Multilingual is released under the cc-by-nc-4.0, which has restrictions for commercial use. Review the license terms before using it in a product.

What's the context length of Jina Reranker v2 Base Multilingual?

Jina Reranker v2 Base Multilingual supports a context window of 1,024 tokens (about 1K).

Source: huggingface.co/jinaai/jina-reranker-v2-base-multilingual

Reviewed by RunLocalAI Editorial. See our editorial policy for how we research and verify model claims.

Related — keep moving

Before you buy

Verify Jina Reranker v2 Base Multilingual runs on your specific hardware before committing money.