List your company

Computer-Using Agent Task on WASP 1.0 (test)

97.6PCRSAFEPRED
53.08864.64476.287.756Feb 2, 2026

Evaluation results

MethodLinks
SAFEPREDFoundation Model=GPT-4o, Guardrail Model=Qwen3-Max2026.0297.697.696.4—————
HarmonyGuardFoundation Model=GPT-4o, Guardrail Model=Qwen3-Max2026.0292.976.267.9—————
Rule-traversedFoundation Model=GPT-4o2026.0289.361.958.3—————
Generic-defenseFoundation Model=GPT-4o2026.0277.454.854.8—————
NoneFoundation Model=GPT-4o2026.0254.863.151.2—————