Skip to content

Benchmark

Vals SkillsBench

Coding benchmarks compare models on programming tasks. Scores are percentages.

Models measured
33
Scores recorded
33
Sources
1
IQ weight
2.21

Score distribution

51.1%
Mean
52.7%
Median
11.1%
Std dev
38.1%
10th percentile
62.2%
90th percentile
18.1%
Min
18.1%69.8%
The band covers the middle 80% of scores. The line marks the median.

This benchmark has a weight of 2.21 for chat models, measured over 33 models. The weight is how much it counts toward llminfo IQ. We measure it from the data, not from the raw scores.

Leaderboard