Skip to content

Benchmark

MTEB(Multilingual, v1)

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

Models measured
20
Scores recorded
20
Sources
1
IQ weight
0.51

Score distribution

55.8%
Mean
58.4%
Median
10%
Std dev
44.2%
10th percentile
68.3%
90th percentile
41.4%
Min
41.4%70.6%
The band covers the middle 80% of scores. The line marks the median.

This benchmark has a weight of 0.51 for embedding models, measured over 20 models. The weight is how much it counts toward llminfo IQ. We measure it from the data, not from the raw scores.

Leaderboard