Skip to content

Benchmark

CyBench Subtasks

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

Models measured
7
Scores recorded
7
Sources
1
IQ weight
0.58

Score distribution

29.1%
Mean
28.7%
Median
12.9%
Std dev
15.2%
10th percentile
43.9%
90th percentile
11.7%
Min
11.7%46.8%
The band covers the middle 80% of scores. The line marks the median.

This benchmark has a weight of 0.58 for chat models, measured over 7 models. The weight is how much it counts toward llminfo IQ. We measure it from the data, not from the raw scores.

Leaderboard

Data credits: CyBench — cite Zhang et al. 2025