Safety Evaluation
Benchmarks
Dataset NameSOTA methodMetricTrendResultsLast Updated
100Refusal Rate
25
Jul 1, 2026
90Refusal Rate
25
Jul 1, 2026
0.1Refusal Rate
25
Jul 1, 2026
32.84Average ASR
25
Jun 8, 2026
30.66Attack Success Rate (ASR)
25
Jun 8, 2026
78.4Accuracy
25
May 29, 2026
81.2Avg Score
25
Feb 26, 2026
8.03ASR
24
Mar 31, 2026
9.13ASR
24
Mar 31, 2026
96.1SFR
24
Mar 18, 2026
0ASR
24
Mar 4, 2026
0.6ASR
24
Mar 4, 2026
88.85Safety Score
24
Mar 4, 2026
99.94CATQA Score
24
Feb 26, 2026
10VSScore
23
Jun 5, 2026
62H Score
22
Jun 1, 2026
0.897Safety Score
22
Mar 25, 2026
3ASR
22
Apr 2, 2026
0.036H.R.adv
22
Feb 26, 2026
95.7Safety Score
21
Jun 11, 2026
13.1S1 ASR (%)
21
May 7, 2026
8.97S1 ASR
21
May 7, 2026
1.27ASR (S1)
21
May 7, 2026
97Safety Score
21
May 21, 2026
86.7DSR (w/ vision)
21
Mar 17, 2026