Jailbreaking
Benchmarks
Dataset NameSOTA methodMetricTrendResultsLast Updated
98.5Bypass Success Rate (Text)
22
Mar 12, 2026
0Jailbreak Success Rate
20
Jun 16, 2026
0ASR (Detoxify)
20
Apr 14, 2026
0ASR (Detoxify)
20
Apr 14, 2026
0ASR (Detoxify)
20
Apr 14, 2026
0.99ASR
19
Mar 25, 2026
98.1ASR@5 (Standard)
19
Feb 26, 2026
96.7ASR (%)
18
Jul 3, 2026
16ASR
18
Jul 3, 2026
93ASR
18
Jul 3, 2026
0.488Average Diversity
15
May 6, 2026
0ASR (Claude-Sonnet-4.5)
14
May 19, 2026
58.7Average Success Rate
14
Apr 14, 2026
96.5ASR
13
Apr 21, 2026
73.3WASR
12
May 19, 2026
0ASR (%)
12
May 18, 2026
69.9ASR
12
May 18, 2026
0.4ASR
12
May 18, 2026
43.2Attack Success Rate (ASR)
12
Mar 27, 2026
88Llama 3 8B ASR
11
May 12, 2026
89.5ASR@1 (Qwen2.5-7B-IT)
11
May 12, 2026
94.55ASR@1 (Qwen2.5-7B-IT)
11
May 12, 2026
4.8BERT Score
11
Mar 11, 2026
0ASR@1 (No Refusal)
11
Feb 26, 2026
76StrongREJECT
10
Jul 7, 2026