Qwen 3 Embedding 4B

Jun 2025
Qwen 3 Embedding 4B
Compare in
on
$0.02per 1M tokens
68.5MTEB English v2
Jun 2025released
2,560Dimensions
33K tokensMax input

Rankingconfiguration unstated?

Full ranking →
1Voyage 3 M Exp68.12Conan Embedding v266.43NV Embed v263.24Qwen 3 Embedding 8B62.85Bge En Icl62.2
···
7Inf Retriever v161.78Qwen 3 Embedding 4B61.69LENS D400060.8

Strong at

S
Embeddings & Retrieval#5 of 200

Alternatives?

ModelScoreCostPriceWhat changes
CheaperQwen 3 Embedding 8BunstatedOpen+2%-50%$0.01Memory (4-bit): 2.4 GB → 4.5 GBVBetterVoyage 3 M Expunstated+11%——Max input: 33K tokens → 32K tokensOpenNV Embed v2unstated+3%——Memory (4-bit): 2.4 GB → 4.7 GBCheaperBge En IclunstatedOpensame-50%$0.01Memory (4-bit): 2.4 GB → 4.3 GBCheaperQwen 3 Embedding 0.6BunstatedOpen-10%-50%$0.01

Run it yourself

estimate
Fits on✓ 8 GB (phone, laptop) · 8-bit✓ 16 GB · full quality✓ 24 GB GPU · full quality✓ 48 GB GPU · full quality✓ 80 GB GPU · full quality
LicenceApache-2.0free for commercial use

Benchmarks

in Embeddings & Retrieval
BEIR · Mean over tasks · English61.6#8
BEIR · Mean over task types · English61.6#8
BEIR · Retrieval · English61.6#8
MTEB Mandarin Chinese v1 · Mean over tasks · Mandarin Chinese72.3#11
MTEB Mandarin Chinese v1 · Mean over task types · Mandarin Chinese73.5#10
MTEB Mandarin Chinese v1 · Retrieval · Mandarin Chinese77#11
MTEB Mandarin Chinese v1 · STS · Mandarin Chinese61.3#23
MTEB Mandarin Chinese v1 · Reranking · Mandarin Chinese66.1#40
MTEB Mandarin Chinese v1 · Clustering · Mandarin Chinese77.9#4
MTEB Mandarin Chinese v1 · Pair Classification · Mandarin Chinese83.3#18
MTEB Mandarin Chinese v1 · Classification · Mandarin Chinese75.5#15
MTEB Europe v1 · Bitext Mining · multilingual (28 languages)92.2#6
MTEB Europe v1 · Clustering · multilingual (29 languages)50.4#10
MTEB Europe v1 · Instruction Reranking · English11.6#2
MTEB Europe v1 · Reranking · multilingual (12 languages)62#14
MTEB Europe v1 · Multilabel Classification · multilingual (23 languages)23.6#9
MTEB French v1 · Pair Classification · French72.6#6
MTEB Indic v1 · Reranking · multilingual (2 languages)85.8#36
MTEB Indic v1 · Pair Classification · Hindi85.6#5
MTEB Indic v1 · STS · multilingual (13 languages)60.8#9
MTEB Indic v1 · Clustering · multilingual (21 languages)49.7#19
MTEB Korean v2 · Mean over tasks · multilingual (2 languages)76.9#2
MTEB Korean v2 · Mean over task types · multilingual (2 languages)74.7#2
MTEB Korean v2 · STS · Korean89.9#1
MTEB Korean v2 · Retrieval · multilingual (2 languages)77.3#4
MTEB Korean v2 · Pair Classification · Korean75.3#2
MTEB Korean v2 · Reranking · Korean67.3#7
MTEB Korean v2 · Classification · Korean67.4#6
MTEB Korean v2 · Clustering · Korean70.9#7
MTEB Dutch v1 · Mean over tasks · Dutch63.7#9
MTEB Dutch v1 · Mean over task types · Dutch68.5#7
MTEB Dutch v1 · Classification · Dutch70.7#10
MTEB Dutch v1 · Clustering · Dutch46.5#12
MTEB Dutch v1 · Retrieval · Dutch65.1#12
MTEB Dutch v1 · STS · Dutch86.1#6
MTEB Dutch v1 · Multilabel Classification · Dutch40.2#9
MTEB Dutch v1 · Pair Classification · Dutch81.5#9
MTEB Dutch v1 · Reranking · Dutch89.1#44
MTEB Portuguese v1 · Multilabel Classification · Portuguese20.6#64
MTEB Medical v1 · Clustering · English44.6#29
MTEB Medical v1 · Reranking · Mandarin Chinese85.1#35
FollowIR · Mean over tasks · English11.6#2
FollowIR · Mean over task types · English11.6#2
FollowIR · Instruction Reranking · English11.6#2
MTEB Spanish v1 · Pair Classification · Spanish80.1#7
MTEB Thai v1 · Clustering · Thai54.3#31
MTEB Thai v1 · Classification · Thai72.4#10
MTEB Thai v1 · Pair Classification · Thai86.7#5
MTEB Thai v1 · Reranking · Thai61.7#12
MTEB Scandinavian v1 · Classification · multilingual (6 languages)72.9#5
MTEB Scandinavian v1 · Retrieval · multilingual (3 languages)68.4#16
MTEB Scandinavian v1 · Bitext Mining · multilingual (3 languages)82.2#10
CoIR · Mean over tasks · multilingual (9 languages)79.2#6
CoIR · Mean over task types · multilingual (9 languages)79.2#6
CoIR · Retrieval · multilingual (9 languages)79.2#6
CoREB v1 · Mean over tasks · multilingual (6 languages)56.5#14
CoREB v1 · Mean over task types · multilingual (6 languages)56.5#14
CoREB v1 · Reranking · multilingual (6 languages)52.3#16
CoREB v1 · Retrieval · multilingual (6 languages)60.6#14
MTEB Code v1 · Mean over tasks · multilingual (15 languages)80.1#7
MTEB Code v1 · Mean over task types · multilingual (15 languages)80.1#7
MTEB Code v1 · Retrieval · multilingual (15 languages)80.1#7
MTEB Russian v1.1 · Mean over tasks · Russian70.1#9
MTEB Russian v1.1 · Mean over task types · Russian67.5#8
MTEB Russian v1.1 · Classification · Russian73.6#11
MTEB Russian v1.1 · STS · Russian78.5#2
MTEB Russian v1.1 · Reranking · Russian68.7#19
MTEB Russian v1.1 · Clustering · Russian65.1#8
MTEB Russian v1.1 · Pair Classification · Russian66.6#11
MTEB Russian v1.1 · Retrieval · Russian76.4#6
MTEB Russian v1.1 · Multilabel Classification · Russian43.3#16
MTEB multilingual v2 · Mean over tasks · multilingual (1037 languages)69.4#6
MTEB multilingual v2 · Mean over task types · multilingual (1037 languages)60.9#6
MTEB multilingual v2 · Classification · multilingual (119 languages)72.3#8
MTEB multilingual v2 · Clustering · multilingual (208 languages)57.1#7
MTEB multilingual v2 · STS · multilingual (35 languages)80.9#3
MTEB multilingual v2 · Pair Classification · multilingual (24 languages)85.1#6
MTEB multilingual v2 · Retrieval · multilingual (120 languages)69.6#9
MTEB multilingual v2 · Bitext Mining · multilingual (1026 languages)79.4#10
MTEB multilingual v2 · Instruction Reranking · English11.6#2
MTEB multilingual v2 · Multilabel Classification · multilingual (25 languages)26.8#15
MTEB multilingual v2 · Reranking · multilingual (22 languages)65.1#22

FAQ

How much does Qwen 3 Embedding 4B cost?›

$0.02 per 1M tokens.

Is Qwen 3 Embedding 4B free to use?›

There is no free API; the cheapest paid price is $0.02 per 1M input tokens. The weights are open, so you can run it yourself.

How good is Qwen 3 Embedding 4B, and what is it best at?›

#5 of 200 in Embeddings. Strongest at: Embeddings & Retrieval.

Is Qwen 3 Embedding 4B better than Inf Retriever v1?›

On BEIR · Mean over tasks · English, Qwen 3 Embedding 4B scores 61.6 vs 61.7 — about the same.

How can I use Qwen 3 Embedding 4B?›

Through the Alibaba (Qwen) API.

When was Qwen 3 Embedding 4B released? Is there a newer version?›

Released 2025-06-05. It is the current version in its line.

Is Qwen 3 Embedding 4B open source?›

Yes, the weights are open (Apache-2.0 licence).

Can I run Qwen 3 Embedding 4B locally?›

The weights are open and can be self-hosted; hardware needs are not confirmed yet.

What is Qwen 3 Embedding 4B?›

Qwen 3 Embedding 4B is an open-weights embeddings & retrieval model from Alibaba (Qwen), released Jun 2025.

What are the best alternatives to Qwen 3 Embedding 4B?›

Qwen 3 Embedding 8B (cheaper), Voyage 3 M Exp (better), NV Embed v2 (open weights).