List your company

Safety on Safety Evaluation Suite (Salad-Bench, WildJailbreak, JailbreakBench, WildChat, WildGuard)

100Safety Rate (S.R.)Agent
68.103276.384184.66592.9459Feb 14, 2026

Evaluation results

MethodLinks
AgentModel Size=Qwen3-4B2026.0210081.3510090.5392.0492.78
STAR-1Model Size=Qwen3-4B2026.0210076.1510083.991.1190.23
BCModel Size=Qwen3-4B2026.0210087.2510096.2296.4295.98
IFPOModel Size=Qwen3-4B2026.0210096.2510099.0499.0798.87
AgentModel Size=Qwen3-8B2026.0210090.810096.8595.3696.6
STAR-1Model Size=Qwen3-8B2026.0210078.110083.5894.391.2
STAR-1-mixModel Size=Qwen3-8B2026.0210062.110074.618784.74
BCModel Size=Qwen3-8B2026.0210091.0510097.2197.7597.2
IFPOModel Size=Qwen3-8B2026.0210096.910099.0598.9498.98
ShortModel Size=Qwen3-14B2026.0210070.9510084.7487.6788.67
AgentModel Size=Qwen3-14B2026.0210094.1510097.5196.8297.7
STAR-1-mixModel Size=Qwen3-14B2026.021007210080.8989.5288.48
BCModel Size=Qwen3-14B2026.0210092.7510097.7497.6197.62
IFPOModel Size=Qwen3-14B2026.0210095.4510099.2298.5498.64
ShortModel Size=Qwen3-4B2026.0299.6846.0510063.6779.8477.85
ShortModel Size=Qwen3-8B2026.0299.6860.2510074.5486.684.21
STAR-1Model Size=Qwen3-14B2026.0299.6883.510083.7495.8992.56
STAR-1-mixModel Size=Qwen3-4B2026.0299.3656.310075.584.3583.1
PureModel Size=Qwen3-14B2026.0299.0440.659755.2377.1973.82
PureModel Size=Qwen3-8B2026.0298.7232.59743.5872.6868.9
PureModel Size=Qwen3-4B2026.0296.4927.359838.8367.6465.66
SafeChainModel Size=Qwen3-8B2026.0275.447.558881.3977.4571.76
SafeChainModel Size=Qwen3-4B2026.0269.3340.67175.2672.9465.83
SafeChainModel Size=Qwen3-14B2026.0269.3340.67175.2672.9465.83