Skip to content

Benchmark

Cybersecurity CTFs

Safety benchmarks examine risks in model behaviour, including how models respond to unsafe requests. Scores are percentages.

Models measured
3
Scores recorded
3
Sources
1
IQ weight
No weight published

Score distribution

51.1%
Mean
46.9%
Median
20.2%
Std dev
28.7%
10th percentile
77.6%
90th percentile
28.7%
Min
28.7%77.6%
The band covers the middle 80% of scores. The line marks the median.

Leaderboard

#ModelScore
1GPT-5.3 Codex· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim77.6%
2Claude Haiku 4.5· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim46.9%
3O1 Mini· LLM Stats data· Reported by the model's developer. The source relaying it did not verify it. claim28.7%

Data credits: Data by LLM Stats