Safety Evaluation
Benchmarks
Dataset NameSOTA methodMetricTrendResultsLast Updated
99.67Safety Score (1-ASR)
8
Feb 26, 2026
100Safety Score
8
Feb 26, 2026
1Safety Score (1-ASR)
8
Feb 26, 2026
1Safety Score (1-ASR)
8
Feb 26, 2026
0.998Safety Score (1-ASR)
8
Feb 26, 2026
1Safety Score (1-ASR)
8
Feb 26, 2026
99.75Safety Score (1-ASR)
8
Feb 26, 2026
97.69Rule Adherence
8
Feb 26, 2026
97.31Rule Score
8
Feb 26, 2026
97.5Rule Adherence Score
8
Feb 26, 2026
97.88Rule Score
8
Feb 26, 2026
2.05HS Rate
8
Feb 26, 2026
2.3Unsafe Ratio
8
Feb 26, 2026
5.3Unsafe Ratio
8
Feb 26, 2026
15.3Chinese Safety Score
8
Feb 26, 2026
19.17Safety Score (Chinese)
8
Feb 26, 2026
43.75Chinese Accuracy
8
Feb 26, 2026
87.5F1 Score
8
Feb 26, 2026
77.9F1 Score
8
Feb 26, 2026
92Compliance (Safe)
7
Jun 11, 2026
0Attack Success Rate (WildJailbreak)
7
Jun 9, 2026
0Attack Success Rate (AD)
7
Jun 9, 2026
91.36ASR
7
May 28, 2026
72.19Accuracy
7
May 19, 2026
96.85Accuracy
7
May 19, 2026