List your company

Multi-turn Safety Risk Assessment on Multi-turn Safety 100 random sampled tasks

0.15ASRToolShield
0.12080.31790.5150.7121Feb 13, 2026

Evaluation results

MethodLinks
ToolShieldModel=Claude-3.5-Sonnet2026.020.150.83
BaselineModel=Claude-3.5-Sonnet2026.020.420.5
ToolShieldModel=Gemini-1.5-Flash2026.020.570.31
BaselineModel=Gemini-1.5-Flash2026.020.690.16
FirewallModel=Gemini-1.5-Flash2026.020.790.11
w/o DefenseModel=Gemini-1.5-Flash2026.020.810.08
FirewallModel=Claude-3.5-Sonnet2026.020.850.13
w/o DefenseModel=Claude-3.5-Sonnet2026.020.880.1