Cohere Rerank 3.5

Advanced reranking model with improved reasoning and multilingual capabilities for enterprise data. Excels in complex enterprise scenarios including Finance, E-commerce, and Hospitality across 100+ languages. If you want to compare the best rerankers for your data, try Agentset.

Leaderboard Rank

#10 of 12

ELO Rating

1451

#10

Win Rate

40.9%

#10

Accuracy (nDCG@10)

0.080

#11

Latency

392ms

#4

Model Information

  • Provider: Cohere
  • License: Proprietary
  • Price per 1M tokens: $0.050
  • Release Date: 2024-12-02
  • Model Name: cohere-rerank-v3.5
  • Total Evaluations: 3300

Performance Record

  • Wins: 1350 (40.9%)
  • Losses: 1867 (56.6%)
  • Ties: 83 (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.

Performance Overview

ELO ratings by dataset

Cohere Rerank 3.5's ELO performance varies across different benchmark datasets, showing its strengths in specific domains.

Cohere Rerank 3.5 - ELO by Dataset

  • arguana: 1224
  • MSMARCO: 1374
  • DBPedia: 1524
  • PG: 1674
  • FiQa: 1692
  • business reports

Detailed Metrics

Dataset breakdown

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

arguana

  • ELO: 169
  • Win Rate: 47.4%
  • Records: 261W-289L-0T

Accuracy Metrics

  • nDCG@50: 0.267
  • nDCG@100: 0.355
  • Recall@50: 0.520
  • Recall@100: 0.800

Latency Distribution

  • Mean: 570ms
  • P50 (Median): 373ms
  • P90: 617ms

MSMARCO

  • ELO: 156
  • Win Rate: 39.5%
  • Records: 217W-295L-38T

Accuracy Metrics

  • nDCG@50: 0.000
  • nDCG@100: 0.000
  • Recall@50: 0.000
  • Recall@100: 0.000

Latency Distribution

  • Mean: 339ms
  • P50 (Median): 285ms
  • P90: 304ms

DBPedia

  • ELO: 147
  • Win Rate: 38.9%
  • Records: 214W-306L-30T

Accuracy Metrics

  • nDCG@50: 0.000
  • nDCG@100: 0.000
  • Recall@50: 0.000
  • Recall@100: 0.000

Latency Distribution

  • Mean: 286ms
  • P50 (Median): 279ms
  • P90: 290ms

PG

  • ELO: 145
  • Win Rate: 47.6%
  • Records: 262W-288L-0T

Accuracy Metrics

  • nDCG@50: 0.000
  • nDCG@100: 0.000
  • Recall@50: 0.000
  • Recall@100: 0.000

Latency Distribution

  • Mean: 458ms
  • P50 (Median): 360ms
  • P90: 615ms

FiQa

  • ELO: 128
  • Win Rate: 31.6%
  • Records: 174W-369L-7T

Accuracy Metrics

  • nDCG@50: 0.124
  • nDCG@100: 0.128
  • Recall@50: 0.123
  • Recall@100: 0.130

Latency Distribution

  • Mean: 364ms
  • P50 (Median): 315ms
  • P90: 401ms

business reports

  • ELO: 123
  • Win Rate: 40.4%
  • Records: 222W-320L-8T

Accuracy Metrics

  • nDCG@50: 0.000
  • nDCG@100: 0.000
  • Recall@50: 0.000
  • Recall@100: 0.000

Latency Distribution

  • Mean: 334ms
  • P50 (Median): 293ms
  • P90: 503ms

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

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);
}