Jailbreak Attack
Benchmarks
Dataset NameSOTA methodMetricTrendResultsLast Updated
100Attack Success Rate AK
9
Feb 26, 2026
99.7ASR (GPT-3.5-Turbo, k)
8
Jun 9, 2026
4.8Model Averaged ASR (0.8)
8
Jun 4, 2026
89ASR
8
May 28, 2026
0.348Loss
8
May 20, 2026
0.106Loss
8
May 20, 2026
87.5ASR
8
May 19, 2026
73.3ASR
8
May 19, 2026
59ASR
8
May 19, 2026
90ASR
8
May 19, 2026
80.5ASR
8
May 19, 2026
85ASR
8
May 19, 2026
1ASR
8
Apr 15, 2026
56.15ASR
8
Apr 14, 2026
62ASR (Google Banana 2)
8
Apr 8, 2026
2.4MFR
8
Apr 8, 2026
1.6Misclassification Failure Rate (MFR)
8
Apr 8, 2026
98CodeCha
8
Mar 18, 2026
93ASR (GPT-4o)
8
Mar 4, 2026
82.64ASR
8
Feb 26, 2026
0.466Loss
7
May 20, 2026
0.61Hypervolume
7
Mar 23, 2026
0.688Hypervolume
7
Mar 23, 2026
0.756Hypervolume
7
Mar 23, 2026
79Hypervolume
7
Mar 23, 2026