Skip to content

Benchmark

SkillsBench 1.1 · Without Skills

Scores are percentages. Higher is better.

Models and systems measured
21
Scores recorded
24
Sources
1
IQ weight
2.55

Score distribution

31.6%
Mean
33.4%
Median
10.3%
Std dev
16.7%
10th percentile
45.7%
90th percentile
8.8%
Min
8.8%51.5%
The band covers the middle 80% of scores. The line marks the median.

This benchmark has a weight of 2.55 for chat models, measured over 19 models. The weight is how much it counts toward llminfo IQ. We measure it from the data, not from the raw scores.

Leaderboard

#Model or systemScore
1GPT 5.5OpenHands · 87 tasks · 3 trials/task · max reasoning effort · without Skills· Reported by SkillsBench / BenchFlow· via BenchBoard docs51.5%
2GPT 5.5Codex · 87 tasks · 3 trials/task · max reasoning effort · without Skills· Reported by SkillsBench / BenchFlow· via BenchBoard docs46.8%
3Claude Opus 4.8OpenHands · 87 tasks · 3 trials/task · max reasoning effort · without Skills· Reported by SkillsBench / BenchFlow· via BenchBoard docs45.7%
4Claude Opus 4.7Claude Code · 87 tasks · 3 trials/task · max reasoning effort · without Skills· Reported by SkillsBench / BenchFlow· via BenchBoard docs43%
5Claude Opus 4.7OpenHands · 87 tasks · 3 trials/task · max reasoning effort · without Skills· Reported by SkillsBench / BenchFlow· via BenchBoard docs42.1%
6Gemini 3.5 FlashOpenHands · 87 tasks · 3 trials/task · max reasoning effort · without Skills· Reported by SkillsBench / BenchFlow· via BenchBoard docs41.1%
7Gemini 3.1 Pro PreviewGemini CLI · 87 tasks · 3 trials/task · max reasoning effort · without Skills· Reported by SkillsBench / BenchFlow· via BenchBoard docs36%
8Gemini 3 FlashGemini CLI · 87 tasks · 3 trials/task · max reasoning effort · without Skills· Reported by SkillsBench / BenchFlow· via BenchBoard docs34.2%
9Gemini 3.1 Pro PreviewOpenHands · 87 tasks · 3 trials/task · max reasoning effort · without Skills· Reported by SkillsBench / BenchFlow· via BenchBoard docs33.8%
10Claude Opus 4.6Claude Code · 87 tasks · 3 trials/task · max reasoning effort · without Skills· Reported by SkillsBench / BenchFlow· via BenchBoard docs33.7%
11Claude Sonnet 4.6OpenHands · 87 tasks · 3 trials/task · max reasoning effort · without Skills· Reported by SkillsBench / BenchFlow· via BenchBoard docs33.5%
12Kimi K2.6OpenHands · 87 tasks · 3 trials/task · max reasoning effort · without Skills· Reported by SkillsBench / BenchFlow· via BenchBoard docs33.4%
13GLM 5.1OpenHands · 87 tasks · 3 trials/task · max reasoning effort · without Skills· Reported by SkillsBench / BenchFlow· via BenchBoard docs32.7%
14GPT-5.4 miniOpenHands · 87 tasks · 3 trials/task · max reasoning effort · without Skills· Reported by SkillsBench / BenchFlow· via BenchBoard docs29.9%
15GPT 5.2Codex · 87 tasks · 3 trials/task · max reasoning effort · without Skills· Reported by SkillsBench / BenchFlow· via BenchBoard docs29.7%
16MiniMax M3OpenHands · 87 tasks · 3 trials/task · max reasoning effort · without Skills· Reported by SkillsBench / BenchFlow· via BenchBoard docs29.7%
17DeepSeek V4 FlashOpenHands · 87 tasks · 3 trials/task · max reasoning effort · without Skills· Reported by SkillsBench / BenchFlow· via BenchBoard docs27.5%
18Claude Opus 4.5Claude Code · 87 tasks · 3 trials/task · max reasoning effort · without Skills· Reported by SkillsBench / BenchFlow· via BenchBoard docs23.8%
19MiniMax M2.7OpenHands · 87 tasks · 3 trials/task · max reasoning effort · without Skills· Reported by SkillsBench / BenchFlow· via BenchBoard docs18.1%
20Claude Sonnet 4.5Claude Code · 87 tasks · 3 trials/task · max reasoning effort · without Skills· Reported by SkillsBench / BenchFlow· via BenchBoard docs16.7%
21Gemini 3.1 Flash LiteOpenHands · 87 tasks · 3 trials/task · max reasoning effort · without Skills· Reported by SkillsBench / BenchFlow· via BenchBoard docs16%
22Claude Haiku 4.5Claude Code · 87 tasks · 3 trials/task · max reasoning effort · without Skills· Reported by SkillsBench / BenchFlow· via BenchBoard docs8.8%

Not ranked

The source marks these results as descriptive, not capability scores, so we list them by name instead of ranking them.

Model or systemScore
DeepSeek V4 Pro (version unspecified)OpenHands · 87 tasks · 3 trials/task · max reasoning effort · without Skills· Reported by SkillsBench / BenchFlow· via BenchBoard docs26.9%
Grok 4.3OpenHands · 87 tasks · 3 trials/task · max reasoning effort · without Skills· Reported by SkillsBench / BenchFlow· via BenchBoard docs22.8%

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.