Skip to content

Benchmark

MTEB(fas, v2) - PairClassification

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

Models measured
12
Scores recorded
12
Sources
1
IQ weight
1.22

Score distribution

83.3%
Mean
83.4%
Median
2.1%
Std dev
81.8%
10th percentile
86.3%
90th percentile
78.5%
Min
78.5%87%
The band covers the middle 80% of scores. The line marks the median.

This benchmark has a weight of 1.22 for embedding models, measured over 12 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 8BPairClassification· MTEB Leaderboard data87%
2Qwen3 Embedding 4BPairClassification· MTEB Leaderboard data86.3%
3Multilingual E5 Large InstructPairClassification· MTEB Leaderboard data85.5%
4Jina Embeddings v5 Omni NanoPairClassification· MTEB Leaderboard data83.6%
5Jina Embeddings v5 Text SmallPairClassification· MTEB Leaderboard data83.4%
6Jina Embeddings v5 Omni SmallPairClassification· MTEB Leaderboard data83.4%
7BGE M3PairClassification· MTEB Leaderboard data83.2%
8E5 Mistral 7B InstructPairClassification· MTEB Leaderboard data83.1%
9Multilingual E5 LargePairClassification· MTEB Leaderboard data82.3%
10Qwen3 Embedding 0.6BPairClassification· MTEB Leaderboard data82.1%
11Jina Embeddings v3PairClassification· MTEB Leaderboard data81.8%
12Text Embedding 3 SmallPairClassification· MTEB Leaderboard data78.5%