Skip to content

Benchmark

SkillsBench · Text-only (79 tasks) · OpenHands

Scores are percentages. Higher is better.

Models measured
6
Scores recorded
6
Sources
1
IQ weight
0.00

Score distribution

62.2%
Mean
63.7%
Median
4.8%
Std dev
62.5%
10th percentile
66.7%
90th percentile
51.9%
Min
51.9%66.7%
The band covers the middle 80% of scores. The line marks the median.

This benchmark has a weight of 0.00 for chat models, measured over 5 models. The weight is how much it counts toward llminfo IQ. We measure it from the data, not from the raw scores.

Leaderboard

#ModelScore
1Qwen3.8 MaxAtria Table 1 / Appendix B · OpenHands within AgentCompass · 79 self-contained text-only tasks · three-run average · reasoning tiers and task-source commit not stated· Reported by Shanghai AI Laboratory· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.66.7%
2DeepSeek V4 Pro 0813Atria Table 1 / Appendix B · OpenHands within AgentCompass · 79 self-contained text-only tasks · three-run average · reasoning tiers and task-source commit not stated· Reported by Shanghai AI Laboratory· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.65%
3Claude Opus 5Atria Table 1 / Appendix B · OpenHands within AgentCompass · 79 self-contained text-only tasks · three-run average · reasoning tiers and task-source commit not stated· Reported by Shanghai AI Laboratory· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.63.7%
4GLM 5.3Atria Table 1 / Appendix B · OpenHands within AgentCompass · 79 self-contained text-only tasks · three-run average · reasoning tiers and task-source commit not stated· Reported by Shanghai AI Laboratory· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.63.3%
5GPT-5.6 SolAtria Table 1 / Appendix B · OpenHands within AgentCompass · 79 self-contained text-only tasks · three-run average · reasoning tiers and task-source commit not stated· Reported by Shanghai AI Laboratory· via BenchBoard docs· Reported by the model's developer. The source relaying it did not verify it.62.5%
6Kimi K3Atria Table 1 / Appendix B · OpenHands within AgentCompass · 79 self-contained text-only tasks · three-run average · reasoning tiers and task-source commit not stated· Reported by Shanghai AI Laboratory· 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.