Skip to content

Benchmark

SuperCLUE 精确指令遵循

Scores are percentages.

Models measured
20
Scores recorded
20
Sources
1
IQ weight
1.85

Score distribution

26.8%
Mean
20.7%
Median
18.8%
Std dev
8.6%
10th percentile
50%
90th percentile
5.2%
Min
5.2%69%
The band covers the middle 80% of scores. The line marks the median.

This benchmark has a weight of 1.85 for chat 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