List your company

Jailbreak Detection on OKT

1Correlation ScoreLlama-Prompt-Guard-2
0.6533680.7433590.833350.923341Feb 14, 2026

Evaluation results

MethodLinks
Llama-Prompt-Guard-2Parameters=86M2026.021
AISABackbone=GPT-OSS-20b-I2026.020.9933
AISABackbone=Llama3.1-8b-I2026.020.99
AISABackbone=Llama2-13b-I2026.020.99
AISABackbone=Mistral-7b-I2026.020.98
AISABackbone=Qwen3-8b-I2026.020.98
GradSafe2026.020.9733
gpt-4.1-miniVersion=2025-04-142026.020.97
gpt-5-miniVersion=2025-08-072026.020.9667
gpt-4o-miniVersion=2024-07-182026.020.9467
Jailbreak-Classifier2026.020.88
NemoGuard-JailbreakDetect2026.020.8633
SPDetector2026.020.6667