List your company

Adversarial Attack Detection on InTheWild

94RecallLlama Prompt Guard 2
8.7230.865375.14Dec 23, 2025

Evaluation results

MethodLinks
Llama Prompt Guard 2Model Size=0.086B, Reasoning=false2025.1294
IBM Granite GuardianVersion=3.1, Model Size=2B, Reasoning=false2025.1287
Apriel GuardModel Size=8B, Reasoning=false2025.1287
Apriel GuardModel Size=8B, Reasoning=true2025.1286
Qwen3GuardModel Size=8B, Constraint=strict, Reasoning=false2025.1285
IBM Granite GuardianVersion=3.2, Model Size=3B, Reasoning=false2025.1280
IBM Granite GuardianVersion=3.2, Model Size=5B, Reasoning=false2025.1279
IBM Granite GuardianVersion=3.3, Model Size=8B, Reasoning=false2025.1277
IBM Granite GuardianVersion=3.3, Model Size=8B, Reasoning=true2025.1269
Qwen3GuardModel Size=8B, Constraint=loose, Reasoning=false2025.1267
gpt-oss-safeguardModel Size=20B, Reasoning=true2025.1261
Llama GuardVersion=4, Reasoning=false2025.1243
Llama GuardVersion=3, Reasoning=false2025.1227
ShieldGemmaModel Size=9B, Reasoning=false2025.1221
Llama GuardVersion=2, Reasoning=false2025.1212