Safety Classification
Benchmarks
Dataset NameSOTA methodMetricTrendResultsLast Updated
37.7F1 Score
64
May 27, 2026
0.783AUROC
49
May 4, 2026
80.5AUROC
49
May 4, 2026
0.94F1 Score
48
Apr 21, 2026
95.3F1 Score
47
Jun 1, 2026
95.8F1 Score
46
Jul 8, 2026
97.3Accuracy
43
May 26, 2026
88.8F1 Score
35
Jun 26, 2026
0.81F1 Score
32
Jun 26, 2026
86.2F1 Score
30
Jul 8, 2026
89.54F1 Score
29
Jun 23, 2026
94AUC
24
May 4, 2026
94AUC
24
May 4, 2026
74.88Accuracy
23
May 26, 2026
8.4ECE
21
May 4, 2026
0.061ECE
21
May 4, 2026
94.01Accuracy
20
Jun 2, 2026
92.04Accuracy
20
Jun 2, 2026
92.91F1
20
Jul 7, 2026
81.4F1 Score
18
Jun 26, 2026
77.93Overall Average Score
16
Jun 16, 2026
85.11Beavertails Score
16
Jun 16, 2026
86.31PolyG
16
Jun 16, 2026
0.934F1 Score
15
May 4, 2026
100Recall
14
Feb 26, 2026