List your company

Adversarial Attacks Detection on Agentic Workflow Aggregate (test)

100PrecisionQwen3Guard-8B-loose
87.5290.769497.24Dec 23, 2025

Evaluation results

MethodLinks
Qwen3Guard-8B-looseReasoning=Without2025.1210033500
Qwen3Guard-8B-strictReasoning=Without2025.129941580
gpt-oss-safeguard-20BReasoning=With2025.129944611
Llama-guard-3-8BReasoning=Without2025.129437534
Llama-guard-4-12BReasoning=Without2025.129448636
AprielGuard-8BReasoning=Without2025.1294929311
AprielGuard-8BReasoning=With2025.1294919211
IBM-Granite-Guardian-3.3-8BReasoning=Without2025.1293818712
Llama-guard-2-8BReasoning=Without2025.129223374
IBM-Granite-Guardian-3.3-8BReasoning=With2025.1292697912
IBM-Granite-Guardian-3.2-5BReasoning=Without2025.1291818616
Llama-Prompt-Guard-2-86MReasoning=Without2025.1290778317
IBM-Granite-Guardian-3.2-3BReasoning=Without2025.1290768216
ShieldGemma-9BReasoning=Without2025.128924386
IBM-Granite-Guardian-3.1-2BReasoning=Without2025.1288878823