Benchmark Almanac

Qwen3-Embedding-4B

Qwen3-Embedding-4B ranks first on DBpedia 14.

47
Benchmarks with a score
1
Ranked first, out of five or more models
13
In the top three
Qwen Jun 2025

Qwen3-Embedding-4B

Open weights Small

No Epoch Capabilities Index score yet
  • CLICK-ID 64.3% #4 /5
  • Estonian Valence Corpus 59.6% #2 /5
  • Greek Legal Code 52.3% #2 /5

Where has it been number one?

Qwen3-Embedding-4B is number one on 1 benchmark today. It also led 2 more before newer models passed it.

Source: Official leaderboards

Where it ranks

Its best score on each benchmark, and where that ranks among every model with a score there. We rank each model once, at its best setting.

Knowledge and language

BenchmarkScoreRankSource
DBpedia 14 99.3% #1 of 5 Official leaderboard
FollowIR 0.12 #1 of 4 Official leaderboard
Norwegian Courts Bitext Mining 93.1% #2 of 6 Official leaderboard
CLIMATE-FEVER 0.47 #1 of 3 Official leaderboard
Ask Ubuntu duplicate question retrieval 0.69 #2 of 5 Official leaderboard
DiaBLa 0.87 #2 of 5 Official leaderboard
Estonian Valence Corpus 59.6% #2 of 5 Official leaderboard
IN22-Gen 82% #2 of 5 Official leaderboard
Amazon Counterfactual Detection 92.9% #2 of 4 Official leaderboard
Corpus for Emotion Detection in Russian 51% #2 of 4 Official leaderboard
CTKFacts 0.89 #2 of 4 Official leaderboard
Faroese Semantic Textual Similarity 0.86 #2 of 4 Official leaderboard
IndicLID benchmark 95% #2 of 4 Official leaderboard
JSICK 88.8% #2 of 4 Official leaderboard
Korean Sarcasm Dataset 61.1% #2 of 4 Official leaderboard
MTEB-BR 0.66 #2 of 4 Official leaderboard
ScaLA 51.3% #2 of 4 Official leaderboard
SemEval 2022 Task 8 multilingual news article similarity 0.72 #2 of 4 Official leaderboard
Tatoeba sentence retrieval 76.3% #3 of 5 Official leaderboard
BRIGHT-Pro 40.8 #2 of 3 Official leaderboard
FarSick 0.74 #2 of 3 Official leaderboard
Q2D-Web 76.5% #2 of 3 Official leaderboard
SemEval 2024 Task 1 semantic textual relatedness 0.63 #3 of 4 Official leaderboard
CLICK-ID 64.3% #4 of 5 Official leaderboard
OBLIQ-Bench 0.05 #4 of 4 Official leaderboard

Math and science

BenchmarkScoreRankSource
ScholarCatalyst 39% #2 of 5 Official leaderboard

Coding

BenchmarkScoreRankSource
CodeSearchNet Challenge 0.92 #2 of 4 Official leaderboard
CodeTransOcean 35.0 #2 of 4 Official leaderboard
CoSQA 0.38 #3 of 4 Official leaderboard

Professional work

BenchmarkScoreRankSource
MultiEURLEX 0.05 #1 of 4 Official leaderboard
AILA Statute Retrieval 0.81 #2 of 5 Official leaderboard
BEIR FiQA-2018 Retrieval 0.63 #2 of 5 Official leaderboard
Financial PhraseBank 93.9% #2 of 5 Official leaderboard
Greek Legal Code 52.3% #2 of 5 Official leaderboard
Bank Question Corpus 0.74 #2 of 4 Official leaderboard
GerDaLIR Small 0.44 #2 of 4 Official leaderboard
GerLayQA 0.24 #2 of 4 Official leaderboard
GerLeRB 0.65 #2 of 4 Official leaderboard
Italian-Administrative-CaseHold 66.7% #2 of 4 Official leaderboard
LegalQuAD 0.69 #2 of 4 Official leaderboard
NFCorpus 0.41 #2 of 4 Official leaderboard
OpenLegalRetrieval-DE 0.42 #2 of 4 Official leaderboard
BANKING77 86.3% #3 of 5 Official leaderboard
GreekBarRetrieval 0.24 #3 of 4 Official leaderboard
BuiltBench 62.5 #2 of 2 Official leaderboard

Other

BenchmarkScoreRankSource
MTEB 71.5% #2 of 4 Official leaderboard
RTEB 67.1% #2 of 2 Official leaderboard

Who measured it?

Independent runs cover all 47 of its benchmarks.

  • An independent run only 47
Official leaderboards scored it on 47.

Sources: Official leaderboards, Epoch AI (CC BY 4.0) and labs’ model cards