List your company

Tool-Using Agent Safety on MT-AgentRisk 1.0 (test)

27.14Filesystem ASRGPT-5.2
24.682841.268957.85574.4411Feb 13, 2026

Evaluation results

MethodLinks
GPT-5.2Setting=Single2026.0227.1471.4322.1467.1425.716022.8671.43604025.4866.3——
Claude-4.5-SonnetSetting=Single2026.0237.1455.7457.8627.1428.5758.5748.5745.712073.344.6643.56——
GPT-5.2Setting=Multi2026.02514034.2950.7137.1454.2931.4367.1486.67039.7350.4155.9-24
Qwen3-CoderSetting=Single2026.0255.7132.8652.143537.1447.148017.1446.6746.6755.0733.97——
Seed-1.6Setting=Single2026.027018.5764.291562.8625.7158.5732.8646.6733.3363.2921.92——
Gemini-3-FlashSetting=Single2026.0271.4324.2965.7115.7148.5741.4378.5714.2953.3333.3365.4822.74——
Seed-1.6Setting=Multi2026.0278.575.7170.717.147018.5778.5711.4360073.159.5915.6-56.3
Deepseek-v3.2Setting=Single2026.02804.29801.435027.1491.432.8673.33076.167.12——
Claude-4.5-SonnetSetting=Multi2026.0282.867.1465.7112.1467.143087.145.7126.676071.7815.4360.7-64.6
Gemini-3-FlashSetting=Multi2026.0282.867.1472.865.7185.714.29808.5780078.96.0320.5-73.5
Qwen3-CoderSetting=Multi2026.0288.571.4367.8613.578010901.4360078.087.6741.8-77.4
Deepseek-v3.2Setting=Multi2026.0288.57080085.714.2992.861.4373.33084.931.111.5-84.5