# Cohere Rerank 4 Fast

Fast cross-encoder reranker for enterprise search and RAG, built for low-latency production workloads. Supports up to 32K context and strong multilingual retrieval across 100+ languages, with optional self-learning to adapt to your domain over time.

## Leaderboard Rank

**Rank:** 7 of 12  
**ELO Rating:** 1510  
**Win Rate:** 49.8%  
**Accuracy (nDCG@10):** 0.094  
**Latency:** 447ms

### Model Information
- **Provider:** Cohere  
- **License:** Proprietary  
- **Price per 1M tokens:** $0.050  
- **Release Date:** 2025-12-11  
- **Model Name:** rerank-v4.0-fast  
- **Total Evaluations:** 3300

### Performance Record
- **Wins:** 1643 (49.8%)  
- **Losses:** 1540 (46.7%)  
- **Ties:** 117 (3.5%)

## 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.

## Performance Overview

### ELO ratings by dataset
Cohere Rerank 4 Fast's ELO performance varies across different benchmark datasets, showing its strengths in specific domains.

#### Cohere Rerank 4 Fast - ELO by Dataset
- business reports: 1426  
- DBPedia: 1471  
- MSMARCO: 1515  
- PG: 1615  
- arguana: 1604  
- FiQa: 1429

## Detailed Metrics

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

### business reports
- **ELO:** 1603  
- **Win Rate:** 56.2%  
- **Record:** 309W-231L-10T

#### Accuracy Metrics
- nDCG@50: 0.000  
- nDCG@100: 0.000  
- Recall@50: 0.000  
- Recall@100: 0.000

#### Latency Distribution
- Mean: 428ms  
- P50 (Median): 408ms  
- P90: 550ms

### DBPedia
- **ELO:** 1580  
- **Win Rate:** 41.4%  
- **Record:** 228W-282L-40T

#### Accuracy Metrics
- nDCG@50: 0.000  
- nDCG@100: 0.000  
- Recall@50: 0.000  
- Recall@100: 0.000

#### Latency Distribution
- Mean: 297ms  
- P50 (Median): 297ms  
- P90: 309ms

### MSMARCO
- **ELO:** 1501  
- **Win Rate:** 45.1%  
- **Record:** 248W-251L-51T

#### Accuracy Metrics
- nDCG@50: 0.000  
- nDCG@100: 0.000  
- Recall@50: 0.000  
- Recall@100: 0.000

#### Latency Distribution
- Mean: 403ms  
- P50 (Median): 382ms  
- P90: 486ms

### PG
- **ELO:** 1474  
- **Win Rate:** 41.6%  
- **Record:** 229W-321L-0T

#### Accuracy Metrics
- nDCG@50: 0.000  
- nDCG@100: 0.000  
- Recall@50: 0.000  
- Recall@100: 0.000

#### Latency Distribution
- Mean: 492ms  
- P50 (Median): 439ms  
- P90: 650ms

### arguana
- **ELO:** 1472  
- **Win Rate:** 62.2%  
- **Record:** 342W-203L-5T

#### Accuracy Metrics
- nDCG@50: 0.351  
- nDCG@100: 0.425  
- Recall@50: 0.660  
- Recall@100: 0.880

#### Latency Distribution
- Mean: 574ms  
- P50 (Median): 562ms  
- P90: 728ms

### FiQa
- **ELO:** 1429  
- **Win Rate:** 42.2%  
- **Record:** 287W-252L-11T

#### Accuracy Metrics
- nDCG@50: 0.135  
- nDCG@100: 0.138  
- Recall@50: 0.125  
- Recall@100: 0.130

#### Latency Distribution
- Mean: 485ms  
- P50 (Median): 459ms  
- P90: 624ms

## 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.

### Code Example

```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 Cohere Rerank 4 Fast stacks up against other top rerankers to understand the differences in performance, accuracy, and latency.