Back to all embeddings

Voyage 3 Large

Produces 2048-dimensional embeddings with flexible dimensionality options (1024, 512, 256) via Matryoshka Representation Learning. Supports binary quantization for up to 200x cost reduction while maintaining retrieval quality.

Leaderboard Rank
#2
of 13
ELO Rating
1528
#2
Win Rate
52.6%
#2
Accuracy (nDCG@10)
0.837
#1
Latency
63477ms
#8

Model Information

Provider
Voyage AI
License
Proprietary
Price per 1M tokens
$0.180
Release Date
2025-01-07
Model Name
voyage-3-large
Total Evaluations
1914

Performance Record

Wins1006 (52.6%)
Losses633 (33.1%)
Ties275 (14.4%)
Wins
Losses
Ties

Performance Overview

ELO ratings by dataset

Voyage 3 Large's ELO performance varies across different benchmark datasets, showing its strengths in specific domains.

Voyage 3 Large - ELO by Dataset

Detailed Metrics

Dataset breakdown

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

SciFact

ELO 156770.4% WR169W-61L-10T

Accuracy Metrics

nDCG@5
0.796
nDCG@10
0.809
Recall@5
0.840
Recall@10
0.880

Latency Distribution

Mean
92317ms
P50 (Median)
90471ms
P90
106165ms

ARCD

ELO 156156.7% WR136W-39L-65T

Accuracy Metrics

nDCG@5
0.953
nDCG@10
0.960
Recall@5
0.960
Recall@10
0.980

Latency Distribution

Mean
36963ms
P50 (Median)
36224ms
P90
42507ms

DBPedia

ELO 154162.8% WR147W-75L-12T

Accuracy Metrics

nDCG@5
0.675
nDCG@10
0.638
Recall@5
0.255
Recall@10
0.362

Latency Distribution

Mean
97066ms
P50 (Median)
95125ms
P90
111626ms

business reports

ELO 152654.6% WR131W-91L-18T

Latency Distribution

Mean
21279ms
P50 (Median)
20853ms
P90
24471ms

FiQa

ELO 151756.3% WR135W-102L-3T

Accuracy Metrics

nDCG@5
0.753
nDCG@10
0.780
Recall@5
0.742
Recall@10
0.837

Latency Distribution

Mean
80239ms
P50 (Median)
78634ms
P90
92275ms

PG

ELO 151755.0% WR132W-106L-2T

Latency Distribution

Mean
93447ms
P50 (Median)
91578ms
P90
107464ms

NorQuAD

ELO 149922.5% WR54W-54L-132T

Latency Distribution

Mean
11003ms
P50 (Median)
10783ms
P90
12653ms

MSMARCO

ELO 149942.5% WR102W-105L-33T

Accuracy Metrics

nDCG@5
0.997
nDCG@10
0.998
Recall@5
0.122
Recall@10
0.222

Latency Distribution

Mean
75502ms
P50 (Median)
73992ms
P90
86827ms

Compare Models

See how it stacks up

Compare Voyage 3 Large with other top embeddings to understand the differences in performance, accuracy, and latency.