List your company

Jailbreak Detection on HB

100Correctness Rate (COR)gpt-4o-mini
16.838.46081.6Feb 14, 2026

Evaluation results

MethodLinks
gpt-4o-miniVersion=2024-07-182026.02100
gpt-5-miniVersion=2025-08-072026.02100
AISABackbone=Qwen3-8b-I2026.0299.5
AISABackbone=Mistral-7b-I2026.0299
AISABackbone=Llama2-13b-I2026.0299
AISABackbone=Llama3.1-8b-I2026.0298.5
gpt-4.1-miniVersion=2025-04-142026.0298
Jailbreak-Classifier2026.0298
AISABackbone=GPT-OSS-20b-I2026.0297.5
GradSafe2026.0295.5
NemoGuard-JailbreakDetect2026.0287.5
SPDetector2026.0268
Llama-Prompt-Guard-2Parameters=86M2026.0220