Text Multilingual Embedding 002

May 2024
Text Multilingual Embedding 002
Compare in
on
$0.025per 1M characters
54.7MTEB English v2
May 2024released
2K tokensMax input
en, Esperanto, Estonian +98Languages

Rankingconfiguration unstated?

Full ranking →
1Seed1.6 Embedding 121591.22Arabic English Sts Matryoshka87.93Harrier OSS v1 27B87.64Qwen 3 Embedding 8B86.65Qwen 3 Embedding 4B85.6
···
28Bge M377.529Text Multilingual Embedding 00277.430Qwen 3 Embedding 0.6B77.2

Strong at

B
Embeddings & Retrieval#64 of 200

Alternatives?

ModelScoreCostPriceWhat changes
BetterEmbeddingGemma 300MunstatedOpen+5%—$0.002Languages: not verifiedoOpenArabic English Sts Matryoshkaunstated+14%——Max input: 2K tokens → 1K tokens+1BetterQwen 3 Embedding 8BunstatedOpen+12%—$0.01Languages: not verifiedBetterQwen 3 Embedding 4BunstatedOpen+11%—$0.02Languages: not verifiedBetterJina Embeddings v5 Text SmallunstatedOpen+3%—$0.05Languages: not verified

Benchmarks

in Embeddings & Retrieval
MTEB Europe v1 · Instruction Reranking · English4.1#16
MTEB Europe v1 · Multilabel Classification · multilingual (23 languages)14.9#63
MTEB Europe v1 · Clustering · multilingual (29 languages)42.4#49
MTEB Europe v1 · Reranking · multilingual (12 languages)59#45
MTEB Europe v1 · Bitext Mining · multilingual (28 languages)87.3#42
MTEB French v1 · Pair Classification · French62.3#50
MTEB Indic v1 · STS · multilingual (13 languages)53.6#19
MTEB Indic v1 · Reranking · multilingual (2 languages)85.2#39
MTEB Indic v1 · Pair Classification · Hindi77.4#29
MTEB Indic v1 · Clustering · multilingual (21 languages)39.2#38
MTEB Dutch v1 · Reranking · Dutch88.1#57
MTEB Portuguese v1 · Multilabel Classification · Portuguese21.7#42
MTEB Medical v1 · Clustering · English35.4#120
FollowIR · Mean over tasks · English4.1#16
FollowIR · Mean over task types · English4.1#16
FollowIR · Instruction Reranking · English4.1#16
MTEB Spanish v1 · Pair Classification · Spanish70.2#33
MTEB Thai v1 · Pair Classification · Thai78.2#28
MTEB Thai v1 · Clustering · Thai45.1#50
MTEB Scandinavian v1 · Bitext Mining · multilingual (3 languages)67.9#62
MTEB Russian v1.1 · Pair Classification · Russian60.4#38
RTEB French beta · Mean over tasks · multilingual (2 languages)41.2#68
MTEB multilingual v2 · Mean over tasks · multilingual (1037 languages)62.2#29
MTEB multilingual v2 · Mean over task types · multilingual (1037 languages)54.3#30
MTEB multilingual v2 · Clustering · multilingual (208 languages)47.8#42
MTEB multilingual v2 · STS · multilingual (35 languages)76.1#18
MTEB multilingual v2 · Classification · multilingual (119 languages)64.6#27
MTEB multilingual v2 · Retrieval · multilingual (120 languages)59.7#43
MTEB multilingual v2 · Multilabel Classification · multilingual (25 languages)22.8#41
MTEB multilingual v2 · Bitext Mining · multilingual (1026 languages)70.7#28
MTEB multilingual v2 · Instruction Reranking · English4.1#16
MTEB multilingual v2 · Reranking · multilingual (22 languages)61.2#60
MTEB multilingual v2 · Pair Classification · multilingual (24 languages)81.1#24

FAQ

How much does Text Multilingual Embedding 002 cost?›

$0.025 per 1M characters.

Is Text Multilingual Embedding 002 free to use?›

There is no free API; it costs $0.025 per 1M characters.

How good is Text Multilingual Embedding 002, and what is it best at?›

#64 of 200 in Embeddings.

Is Text Multilingual Embedding 002 better than Bge M3?›

On MTEB Indic v1 · Pair Classification · Hindi, Text Multilingual Embedding 002 scores 77.4 vs 77.5 — about the same. It is 150% more expensive.

When was Text Multilingual Embedding 002 released? Is there a newer version?›

Released 2024-05-14. It is the current version in its line.

Is Text Multilingual Embedding 002 open source?›

No, Text Multilingual Embedding 002 is a closed model available through an API. The closest open-weights option here is Arabic English Sts Matryoshka.

What is Text Multilingual Embedding 002?›

Text Multilingual Embedding 002 is a closed embeddings & retrieval model from Google, released May 2024.

What are the best alternatives to Text Multilingual Embedding 002?›

EmbeddingGemma 300M (better), Arabic English Sts Matryoshka (open weights).