List your company

Prompt Injection Defense on GSM8K PI (test)

3.3ASR@1No Defense
3.0045.00278.998Jan 21, 2026

Evaluation results

MethodLinks
No DefenseBackbone=Qwen3-235B-A22B, Topology=Random2026.013.37.71196.793.390
AgentXposed-KickBackbone=Qwen3-235B-A22B, Topology=Random2026.013.38.710.796.791.790
INFA-GUARDBackbone=Qwen3-235B-A22B, Topology=Random2026.013.33.73.396.796.796.7
G-safeguardBackbone=Qwen3-235B-A22B, Topology=Random2026.013.73.33.396.796.796.7
AgentSafeBackbone=Qwen3-235B-A22B, Topology=Random2026.013.77.310.396.793.390
AgentXposed-GuideBackbone=Qwen3-235B-A22B, Topology=Random2026.014.35.77.3959593.3
ChallengerBackbone=Qwen3-235B-A22B, Topology=Random2026.015.514.515.99384.784.7
No DefenseBackbone=GPT-4o-mini, Topology=Random2026.016.7131893.386.781.7
G-safeguardBackbone=GPT-4o-mini, Topology=Random2026.016.76.76.793.393.393.3
INFA-GUARDBackbone=GPT-4o-mini, Topology=Random2026.016.76.76.793.393.393.3
AgentSafeBackbone=GPT-4o-mini, Topology=Random2026.01721.731.393.38071.7
InspectorBackbone=Qwen3-235B-A22B, Topology=Random2026.017.210.316.192.392.382.1
ChallengerBackbone=GPT-4o-mini, Topology=Random2026.01911.714.288.376.771.7
AgentXposed-GuideBackbone=GPT-4o-mini, Topology=Random2026.019.32026.79083.375
InspectorBackbone=GPT-4o-mini, Topology=Random2026.0110.312.212.488.38583.3
AgentXposed-KickBackbone=GPT-4o-mini, Topology=Random2026.0110.726.335.788.373.365