Safety Evaluation
Benchmarks
Dataset NameSOTA methodMetricTrendResultsLast Updated
1.26Harm Score
15
Jun 16, 2026
2ASR (Raw)
15
May 29, 2026
0ASR
15
May 13, 2026
0ASR
15
May 13, 2026
98Blue Reward (Higher)
15
Mar 12, 2026
95Safety
15
Mar 5, 2026
77.4Safe Rate
15
Feb 26, 2026
6.52Attack Success Rate (ASR)
14
May 28, 2026
7.07ASR
14
May 28, 2026
14.68Attack Success Rate (ASR)
14
May 28, 2026
41.6Avg Success Rate (Reasoning Path)
14
May 7, 2026
93.08F1 Score
14
Jun 23, 2026
0.9ASR
14
Mar 30, 2026
20.4ASR
14
Mar 30, 2026
88.6HarmR
14
Mar 24, 2026
5.9HarmR
14
Mar 24, 2026
4.93Ring-16 Score
13
Feb 26, 2026
69.85WildChat Score
13
Feb 26, 2026
0.38Unsafe Rate
13
Feb 26, 2026
0Unsafe Rate
13
Feb 26, 2026
0Unsafe Rate
13
Feb 26, 2026
0Unsafe Rate
13
Feb 26, 2026
26.9Compliance Rate
12
Jun 9, 2026
88.97Safe Rate
12
Jun 8, 2026
93.8H Score
12
Jun 1, 2026