Back to all embeddings

Voyage 4

Mid-sized model approaching the retrieval quality of voyage-3-large while maintaining mid-tier model efficiency. Part of shared embedding space with Voyage 4 family, enabling asymmetric retrieval. Supports Matryoshka learning (2048/1024/512/256 dims) and multiple quantization options. If you want to compare the best embedding models for your data, try Agentset.

Leaderboard Rank
#3
of 18
ELO Rating
1586
#3
Win Rate
57.0%
#3
Accuracy (nDCG@10)
0.624
#13
Latency
339ms
#17

Model Information

Provider
Voyage AI
License
Proprietary
Price per 1M tokens
$0.060
Dimensions
1024
Release Date
2026-01-15
Model Name
voyage-4
Total Evaluations
1120

Performance Record

Wins639 (57.1%)
Losses383 (34.2%)
Ties98 (8.8%)
Wins
Losses
Ties

Embedding Models 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 embeddings to manage.

Trusted by teams building production RAG applications

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

Performance Overview

ELO ratings by dataset

Voyage 4's ELO performance varies across different benchmark datasets, showing its strengths in specific domains.

Voyage 4 - ELO by Dataset

Detailed Metrics

Dataset breakdown

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

PG

ELO 150044.0% WR22W-27L-1T

Accuracy Metrics

nDCG@5
0.000
nDCG@10
0.000
Recall@5
0.000
Recall@10
0.000

Latency Distribution

Mean
311ms
P50 (Median)
309ms
P90
321ms

business reports

ELO 150058.9% WR106W-67L-7T

Accuracy Metrics

nDCG@5
0.000
nDCG@10
0.000
Recall@5
0.000
Recall@10
0.000

Latency Distribution

Mean
309ms
P50 (Median)
310ms
P90
325ms

DBPedia

ELO 150060.0% WR108W-54L-18T

Accuracy Metrics

nDCG@5
0.815
nDCG@10
0.811
Recall@5
0.062
Recall@10
0.122

Latency Distribution

Mean
327ms
P50 (Median)
312ms
P90
357ms

FiQa

ELO 150062.2% WR112W-66L-2T

Accuracy Metrics

nDCG@5
0.873
nDCG@10
0.859
Recall@5
0.763
Recall@10
0.840

Latency Distribution

Mean
310ms
P50 (Median)
311ms
P90
324ms

SciFact

ELO 150062.2% WR112W-61L-7T

Accuracy Metrics

nDCG@5
0.737
nDCG@10
0.758
Recall@5
0.804
Recall@10
0.878

Latency Distribution

Mean
321ms
P50 (Median)
311ms
P90
331ms

MSMARCO

ELO 150045.0% WR81W-68L-31T

Accuracy Metrics

nDCG@5
0.941
nDCG@10
0.931
Recall@5
0.123
Recall@10
0.221

Latency Distribution

Mean
317ms
P50 (Median)
307ms
P90
323ms

ARCD

ELO 150057.6% WR98W-40L-32T

Accuracy Metrics

nDCG@5
0.916
nDCG@10
0.916
Recall@5
0.980
Recall@10
0.980

Latency Distribution

Mean
477ms
P50 (Median)
310ms
P90
331ms

Build RAG in Minutes, Not Months

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

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 Voyage 4 with other top embeddings to understand the differences in performance, accuracy, and latency.