Skip to content

Benchmark

MTEB(pol, v1) - Clustering

Embedding benchmarks measure how well models find related items, for example when matching a search to relevant results. Scores are percentages.

Models measured
25
Scores recorded
25
Sources
1
IQ weight
0.54

Score distribution

36.8%
Mean
37.7%
Median
9.9%
Std dev
22.6%
10th percentile
49.5%
90th percentile
22.2%
Min
22.2%50.5%
The band covers the middle 80% of scores. The line marks the median.

This benchmark has a weight of 0.54 for embedding models, measured over 25 models. The weight is how much it counts toward llminfo IQ. We measure it from the data, not from the raw scores.

Leaderboard

#ModelScore
1Qwen3 Embedding 8BClustering· MTEB Leaderboard data50.5%
2Jina Embeddings v5 Omni SmallClustering· MTEB Leaderboard data49.6%
3Jina Embeddings v5 Text SmallClustering· MTEB Leaderboard data49.5%
4Multilingual E5 Large InstructClustering· MTEB Leaderboard data49%
5Qwen3 Embedding 4BClustering· MTEB Leaderboard data49%
6Jina Embeddings v5 Omni NanoClustering· MTEB Leaderboard data48.8%
7E5 Mistral 7B InstructClustering· MTEB Leaderboard data48.8%
8BGE multilingual gemma2Clustering· MTEB Leaderboard data46.4%
9Qwen3 Embedding 0.6BClustering· MTEB Leaderboard data44.4%
10Multilingual E5 LargeClustering· MTEB Leaderboard data39%
11snowflake-arctic-embed-l-v2.0Clustering· MTEB Leaderboard data38.8%
12Paraphrase Multilingual MPNet v2Clustering· MTEB Leaderboard data38.3%
13Granite Embedding 278m MultilingualClustering· MTEB Leaderboard data37.7%
14Gte LargeClustering· MTEB Leaderboard data32.3%
15UAE-Large-V1Clustering· MTEB Leaderboard data31.5%
16BGE Large En V1.5Clustering· MTEB Leaderboard data31.1%
17Gte BaseClustering· MTEB Leaderboard data29.8%
18BGE Base En V1.5Clustering· MTEB Leaderboard data29%
19E5 Base V2Clustering· MTEB Leaderboard data27.9%
20bge-small-en-v1.5Clustering· MTEB Leaderboard data27.5%
21E5 Large V2Clustering· MTEB Leaderboard data27.1%
22All Mpnet Base V2Clustering· MTEB Leaderboard data26.5%
23All MiniLM L12 V2Clustering· MTEB Leaderboard data22.6%
24All MiniLM L6 V2Clustering· MTEB Leaderboard data22.3%
25Jina Embedding B EN v1Clustering· MTEB Leaderboard data22.2%