V

Voyage Large 2 Instruct

May 2024
VVoyage Large 2 Instruct
Compare in
on
$0.12per 1M tokens
91.3MTEB English v2
May 2024released
16K tokensMax input

Rankingconfiguration unstated?

Full ranking →
1Jina Embeddings v5 Omni Nano81.82Jina Embeddings v5 Omni Small81.73Jina Embeddings v5 Text Small81.74Qwen 3 Embedding 8B81.65Qwen 3 Embedding 4B81.2
···
29Paraphrase Multilingual Mpnet Base v274.330Voyage Large 2 Instruct7431F2LLM v2 330M74

Alternatives?

ModelScoreCostPriceWhat changes
CheaperJina Embeddings v5 Text SmallunstatedOpen+10%-58%$0.05BetterEmbeddingGemma 300MunstatedOpen+4%-98%$0.002Max input: 16K tokens → 2K tokensCheaperQwen 3 Embedding 8BunstatedOpen+10%-92%$0.01CheaperQwen 3 Embedding 4BunstatedOpen+10%-83%$0.02CheaperJina Embeddings v3unstatedOpen+6%-58%$0.05Max input: 16K tokens → 8K tokens

Benchmarks

in Embeddings & Retrieval
BRIGHT · Mean over tasks · English17.6#9
BRIGHT · Mean over task types · English17.6#9
BRIGHT · Retrieval · English17.6#9
JMTEB v2 · Clustering · Japanese50.5#37
MTEB German v1 · Classification · German70.5#26
MTEB German v1 · Pair Classification · German55.6#20
MTEB Europe v1 · STS · multilingual (8 languages)74#30
MTEB Europe v1 · Pair Classification · multilingual (11 languages)87#53
MTEB Europe v1 · Bitext Mining · multilingual (28 languages)78.2#84
MTEB French v1 · Pair Classification · French64.3#37
MTEB Indic v1 · Bitext Mining · multilingual (20 languages)13.8#65
MTEB Indic v1 · STS · multilingual (13 languages)6.9#118
MTEB Indic v1 · Reranking · multilingual (2 languages)73.1#93
MTEB Indic v1 · Pair Classification · Hindi59.4#100
MTEB Indic v1 · Clustering · multilingual (21 languages)12.4#99
MTEB Dutch v1 · Reranking · Dutch87.7#63
MTEB Polish v1 · Classification · Polish61.9#32
MTEB Polish v1 · Pair Classification · Polish81.2#51
MTEB Portuguese v1 · Multilabel Classification · Portuguese17.2#162
MTEB Medical v1 · Clustering · English39.3#55
MTEB Medical v1 · Reranking · Mandarin Chinese79.6#54
MTEB Spanish v1 · Pair Classification · Spanish64.2#65
MTEB Thai v1 · Pair Classification · Thai60.3#99
MTEB Thai v1 · Clustering · Thai27.3#97
MTEB Scandinavian v1 · Mean over tasks · multilingual (6 languages)59.7#25
MTEB Scandinavian v1 · Mean over task types · multilingual (6 languages)59.1#28
MTEB Scandinavian v1 · Classification · multilingual (6 languages)61.7#32
MTEB Scandinavian v1 · Retrieval · multilingual (3 languages)69.2#18
MTEB Scandinavian v1 · Clustering · multilingual (2 languages)43.3#29
MTEB Scandinavian v1 · Bitext Mining · multilingual (3 languages)62.3#119
MTEB Russian v1.1 · Classification · Russian63.8#33
MTEB Russian v1.1 · Clustering · Russian57.6#35
MTEB Russian v1.1 · Multilabel Classification · Russian38.6#38
MTEB Russian v1.1 · Pair Classification · Russian56.4#73
RTEB French beta · Mean over tasks · multilingual (2 languages)34.2#84
MTEB multilingual v2 · STS · multilingual (35 languages)70.1#51
MTEB multilingual v2 · Bitext Mining · multilingual (1026 languages)46.8#86
MTEB multilingual v2 · Pair Classification · multilingual (24 languages)77.8#57

FAQ

How much does Voyage Large 2 Instruct cost?›

$0.12 per 1M tokens.

Is Voyage Large 2 Instruct free to use?›

There is no free API; it costs $0.12 per 1M tokens.

How good is Voyage Large 2 Instruct, and what is it best at?›

#4 of 238 in Embeddings.

Is Voyage Large 2 Instruct better than Paraphrase Multilingual Mpnet Base v2?›

On MTEB Europe v1 · STS · multilingual (8 languages), Voyage Large 2 Instruct scores 74 vs 74.3 — about the same.

When was Voyage Large 2 Instruct released? Is there a newer version?›

Released 2024-05-05. It is the current version in its line.

Is Voyage Large 2 Instruct open source?›

No, Voyage Large 2 Instruct is a closed model available through an API. The closest open-weights option here is Jina Embeddings v5 Text Small.

What is Voyage Large 2 Instruct?›

Voyage Large 2 Instruct is a closed embeddings & retrieval model from Voyage AI (MongoDB), released May 2024.

What are the best alternatives to Voyage Large 2 Instruct?›

Jina Embeddings v5 Text Small (cheaper), EmbeddingGemma 300M (better).