List your company

Prompt Injection Detection on PopUp attack (top visited websites)

99.98AccuracyWARD-0.8b
42.717657.583872.4587.3162Apr 14, 2026Apr 19, 2026Apr 24, 2026Apr 29, 2026May 4, 2026May 9, 2026May 14, 2026

Evaluation results

MethodLinks
WARD-0.8bModel Category=Ours, Number of Parameters=0.8b2026.0599.9810099.9699.98
WARD-2bModel Category=Ours, Number of Parameters=2b2026.0599.9110099.8199.91
Claude-Sonnet-4.6Model Category=Closed-source API2026.0599.6699.799.6399.66
Gemini-3-FlashModel Category=Closed-source API2026.0598.9699.7898.1798.96
GPT-5.4Model Category=Closed-source API2026.0598.999.5998.2398.91
GPT-5.4-MiniModel Category=Closed-source API2026.0598.2196.6899.7398.18
Claude-Haiku-4.5Model Category=Closed-source API2026.0597.8698.0697.6697.86
WebAgentGuard-8BModel Category=Ours2026.0491.1390.1891.3190.74
Gemini-3.1-proModel Category=Closed-source API2026.0591.0399.4885.191.73
GPT-4.1Model Category=Closed-source APIs2026.0490.4580.0810088.94
Qwen-3.5-4BModel Category=Open-source instructed model, Number of Parameters=4B2026.0588.0977.1898.7186.63
Gemma-4-31BModel Category=Open-source instructed model, Number of Parameters=31B2026.0586.9974.7299.0185.17
Qwen3.5-35B-A3BModel Category=Open-source instructed model, Number of Parameters=35B2026.0586.8975.5897.6985.22
WebAgentGuard-8bModel Category=Guard model (Prompt Injection), Number of Parameters=8b2026.0586.6376.1496.3785.07
Qwen-3.5-2BModel Category=Open-source instructed model, Number of Parameters=2B2026.0584.8684.6585.0284.83
GPT-4oModel Category=Closed-source APIs2026.0482.7878.8699.8688.13
WebAgentGuard-4BModel Category=Ours2026.0482.1170.5589.9879.09
Qwen-3.5-0.8BModel Category=Open-source instructed model, Number of Parameters=0.8B2026.0576.3474.3177.4675.85
Qwen3-VL-Instruct-4BModel Category=Open-source instructed models2026.0472.2144.3797.0360.9
GPT-4o-MiniModel Category=Closed-source APIs2026.0470.0641.5291.7857.18
WebAgentGuard-4bModel Category=Guard model (Prompt Injection), Number of Parameters=4b2026.0566.9337.5591.0553.17
Gemma-3-4BModel Category=Open-source instructed model, Number of Parameters=4B2026.0565.668.9464.6366.71
Qwen3-VL-Instruct-8BModel Category=Open-source instructed models2026.0462.822.6299.0936.83
Qwen2.5-VL-Instruct-7BModel Category=Open-source instructed models2026.0459.5321.3587.4434.32
Prompt-Guard-2-86MModel Category=Guard models2026.0452.052.06503.96
GuardReasoner-VL-7BModel Category=Guard models2026.04521.34502.61
Llama-Guard-3-Vision-11BModel Category=Guard models2026.0451.654.0245.347.39
Prompt-Guard-1-86MModel Category=Guard models2026.0450.8133.1248.1239.24
Llama-Guard-4-12BModel Category=Guard model (Jail Break), Number of Parameters=12B2026.0550.040.2658.330.52
GPT-Oss-Safeguard-20bModel Category=Guard model (Jail Break), Number of Parameters=20b2026.0550000
Prompt-Guard-1-86MModel Category=Guard model (Jail Break), Number of Parameters=86M2026.05501.79503.46
Prompt-Guard-2-86MModel Category=Guard model (Jail Break), Number of Parameters=86M2026.05500.3500.59
Llama-Guard-3-Vision-11BModel Category=Guard model (Jail Break), Number of Parameters=11B2026.0550000
GuardReasoner-VL-7BModel Category=Guard model (Jail Break), Number of Parameters=7B2026.0550000
BrowseSafeModel Category=Guard model (Prompt Injection)2026.0550000
SuperAgent-Guard-1.7bModel Category=Guard model (Prompt Injection), Number of Parameters=1.7b2026.0550000
SuperAgent-Guard-4bModel Category=Guard model (Prompt Injection), Number of Parameters=4b2026.0550000
DataSentinelModel Category=Guard model (Prompt Injection)2026.0549.9648.7749.9649.36
PromptArmorModel Category=Guard model (Prompt Injection)2026.0549.7887.9249.8763.64
Llama-3.2-Vision-Instruct-11BModel Category=Open-source instructed models2026.0444.925.5340.879.74