Skip to content

Benchmark

SkillsBench · Self-contained (79 tasks) · Claude Code

Coding benchmarks compare models on programming tasks. Scores are percentages. Higher is better.

Models measured
4
Scores recorded
4
Sources
1
IQ weight
No weight published

Score distribution

57.7%
Mean
61.5%
Median
5.5%
Std dev
51.9%
10th percentile
64.6%
90th percentile
51.9%
Min
51.9%64.6%
The band covers the middle 80% of scores. The line marks the median.

Leaderboard

#ModelScore
1Claude Opus 4.8Dots Studio official appendix · footnote 11 · Claude Code · 79 tasks excluding external API-dependent tasks · three-run average · reasoning effort not stated· Reported by Dots Studio / Xiaohongshu· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.64.6%
2DeepSeek V4 Flash 0731Dots Studio official appendix · footnote 11 · Claude Code · 79 tasks excluding external API-dependent tasks · three-run average · reasoning effort not stated· Reported by Dots Studio / Xiaohongshu· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.61.5%
3Dots3-Note PreviewDots Studio official appendix · footnote 11 · Claude Code · 79 tasks excluding external API-dependent tasks · three-run average · reasoning effort not stated· Reported by Dots Studio / Xiaohongshu· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.52.8%
4GLM 5.2Dots Studio official appendix · footnote 11 · Claude Code · 79 tasks excluding external API-dependent tasks · three-run average · reasoning effort not stated· Reported by Dots Studio / Xiaohongshu· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.51.9%

Data credits: BenchBoard by SYuan03 (github.com/SYuan03/benchboard). MIT covers its code and data organization; no results licence is inferred. Numeric relays retain the original stating-source links.