List your company

Safety risk detection on TS-Bench

86Overall F1 ScoreBreeze Guard
68.3272.9177.582.09Mar 7, 2026

Evaluation results

MethodLinks
Breeze GuardInference mode=non-thinking2026.0386858087889897
Breeze GuardInference mode=thinking2026.0384937387899395
Granite Guardian 3.3Model scale=8B2026.03691838808986100