Share your thoughts, 1 month free Claude Pro on usSee more
WorkDL logo mark

Jailbreaking on HarmBench Closed-Source Models

86GPT-4o Jailbreak Success Rate

DIALTREE

9.5629.40549.2569.095Oct 2, 2025
Updated 1mo ago

Evaluation Results

MethodLinks
2025.10
869086.587.5757182.7
2025.10
4854.5193410.59.529.3
2025.10
43.5451737.570.525.1
2025.10
40.543.554.538.518833.8
2025.10
2715.59.5530.51.514.8
2025.10
25.53523.5274.52623.6
2025.10
20256.53122.5418.2
2025.10
181611.520.58.52.512.8
2025.10
12.55.5025.5107.4