# Jina Reranker v2 Base Multilingual

Open-source 278M parameter cross-encoder with Flash Attention 2, designed for agentic RAG with function-calling awareness. Ranked highly on AirBench leaderboard at release with support for 100+ languages and code retrieval optimization. If you want to compare the best rerankers for your data, try [Agentset](https://app.agentset.ai/).

## Leaderboard Rank

**#12 of 12**  
**ELO Rating**  
1327  
**Win Rate**  
28.2%  
**Accuracy (nDCG@10)**  
0.080  
**Latency**  
746ms

## Model Information

- **Provider** Jina AI
- **License** cc-by-nc-4.0
- **Price per 1M tokens** $0.045
- **Release Date** 2024-06-25
- **Model Name** jina-reranker-v2-base-multilingual
- **Total Evaluations** 3300

## Performance Record

- **Wins** 931 (28.2%)
- **Losses** 2287 (69.3%)
- **Ties** 82 (2.5%)

## Rerankers Are Just One Piece of RAG

Agentset gives you a managed RAG pipeline with the top-ranked models and best practices baked in. No infrastructure to maintain, no reranking to configure.

Trusted by teams building production RAG applications

- **5M+** Documents  
- **1,500+** Teams  
- **99.9%** Uptime

## Performance Overview

### ELO ratings by dataset

Jina Reranker v2 Base Multilingual's ELO performance varies across different benchmark datasets, showing its strengths in specific domains.

#### Jina Reranker v2 Base Multilingual - ELO by Dataset

| Dataset   | ELO   | Win Rate  |  Wins    | Losses  | Ties  |
|-----------|-------|-----------|----------|---------|-------|
| PG        | 1398  | 28.9%     | 159      | 390     | 1     |
| arguana   | 1396  | 37.5%     | 206      | 344     | 0     |
| MSMARCO   | 1381  | 34.7%     | 191      | 315     | 44    |
| DBPedia   | 1326  | 27.6%     | 152      | 367     | 31    |
| business reports | 1281  | 20.2%     | 111      | 439     | 0     |
| FiQa      | 1181  | 20.4%     | 112      | 432     | 6     |

## Detailed Metrics

### Dataset breakdown

Performance metrics across different benchmark datasets, including accuracy and latency percentiles.

#### PG

- **ELO** 1398
- **Win Rate** 28.9%  
**Record** 159W-390L-1T

##### Accuracy Metrics
- nDCG@50.000  
- nDCG@100.000  
- Recall@50.000  
- Recall@100.000

##### Latency Distribution
- Mean 1059ms  
- P50 (Median) 823ms  
- P90 1744ms

#### arguana

- **ELO** 1396
- **Win Rate** 37.5%  
**Record** 206W-344L-0T

##### Accuracy Metrics
- nDCG@50.314  
- nDCG@100.374  
- Recall@50.580  
- Recall@100.760

##### Latency Distribution
- Mean 689ms  
- P50 (Median) 617ms  
- P90 834ms

#### MSMARCO

- **ELO** 1381
- **Win Rate** 34.7%  
**Record** 191W-315L-44T

##### Accuracy Metrics
- nDCG@50.000  
- nDCG@100.000  
- Recall@50.000  
- Recall@100.000

##### Latency Distribution
- Mean 694ms  
- P50 (Median) 616ms  
- P90 911ms

#### DBPedia

- **ELO** 1326
- **Win Rate** 27.6%  
**Record** 152W-367L-31T

##### Accuracy Metrics
- nDCG@50.000  
- nDCG@100.000  
- Recall@50.000  
- Recall@100.000

##### Latency Distribution
- Mean 671ms  
- P50 (Median) 614ms  
- P90 825ms

#### business reports

- **ELO** 1281
- **Win Rate** 20.2%  
**Record** 111W-439L-0T

##### Accuracy Metrics
- nDCG@50.000  
- nDCG@100.000  
- Recall@50.000  
- Recall@100.000

##### Latency Distribution
- Mean 690ms  
- P50 (Median) 620ms  
- P90 824ms

#### FiQa

- **ELO** 1181
- **Win Rate** 20.4%  
**Record** 112W-432L-6T

##### Accuracy Metrics
- nDCG@50.105  
- nDCG@100.108  
- Recall@50.088  
- Recall@100.093

##### Latency Distribution
- Mean 676ms  
- P50 (Median) 626ms  
- P90 837ms

## Build RAG in Minutes, Not Months

Agentset gives you a complete RAG API with top-ranked rerankers and embedding models built in. Upload your data, call the API, and get accurate results from day one.

### Example Code Snippet

```typescript
import { Agentset } from "agentset";

const agentset = new Agentset();
const ns = agentset.namespace("ns_1234");

const results = await ns.search(
  "What is multi-head attention?"
);

for (const result of results) {
  console.log(result.text);
}
```

## Compare Models

### See how it stacks up

Compare Jina Reranker v2 Base Multilingual with other top rerankers to understand the differences in performance, accuracy, and latency.

- **vs Zerank 2** - ZeroEntropy - ELO 1638 - Win Rate 57.2%  
- **vs Cohere Rerank 4 Pro** - Cohere - ELO 1629 - Win Rate 57.7%  
- **vs Zerank 1** - ZeroEntropy - ELO 1573 - Win Rate 57.2%

[View Full Leaderboard](/content/rerankers/index.html)