List your company

Attack Success Rate on 20,000 harmful requests and 20,000 jailbreak prompts (test)

80.2Attack Success Rate (ASR)GateBreaker
3.44823.37443.363.226Dec 24, 2025

Evaluation results

MethodLinks
GateBreakerTarget Model=GPT-OSS-20B2025.1280.2
GateBreakerTarget Model=Hunyuan-A13B-Instruct2025.1276.9
GateBreakerTarget Model=OpenPangu-Pro-MoE-72B2025.1273.1
GateBreakerTarget Model=Average2025.1264.9
GateBreakerTarget Model=Mixtral-8x7B-Instruct-v0.12025.1264.2
GateBreakerTarget Model=Qwen1.5-MoE-A2.7B-Chat2025.1257.8
GateBreakerTarget Model=Qwen3-30B-A3B-Instruct-25072025.1256.9
GateBreakerTarget Model=Phi-3.5-MoE-Instruct2025.1256.5
GateBreakerTarget Model=Deepseek-MoE-16b-Chat2025.1253.6
SAFExTarget Model=Mixtral-8x7B-Instruct-v0.12025.1248.8
SAFExTarget Model=Deepseek-MoE-16b-Chat2025.1235.6
SAFExTarget Model=Qwen1.5-MoE-A2.7B-Chat2025.1235
SAFExTarget Model=OpenPangu-Pro-MoE-72B2025.1230
SAFExTarget Model=Average2025.1229.9
SAFExTarget Model=Qwen3-30B-A3B-Instruct-25072025.1228.4
SAFExTarget Model=Hunyuan-A13B-Instruct2025.1228.4
SAFExTarget Model=Phi-3.5-MoE-Instruct2025.1226.8
SAFExTarget Model=GPT-OSS-20B2025.126.4