Input Moderation Benchmark Suite (ToxicChat, OAIMod, Aegis, Aegis2, SSTest, HarmB, WildG)
88.2Macro-average F1Llama3-StreamGuard-8B
Evaluation Results
| Method | Links | |||||||||
|---|---|---|---|---|---|---|---|---|---|---|
| Llama3-StreamGuard-8BStreaming=yes2026.04 | 88.2 | — | — | — | — | — | — | — | — | |
| Llama3-StreamGuard-3BStreaming=yes2026.04 | 87.8 | — | — | — | — | — | — | — | — | |
| PolyGuard-Qwen-7BStreaming=no2026.04 | 87 | — | — | — | — | — | — | — | — | |
| Qwen3Guard-Stream-8B-strictStreaming=yes2026.04 | 86.7 | — | — | — | — | — | — | — | — | |
| Llama3-StreamGuard-1BStreaming=yes2026.04 | 86.4 | — | — | — | — | — | — | — | — | |
| Qwen3Guard-8B-Gen-looseStreaming=no2026.04 | 86.3 | — | — | — | — | — | — | — | — | |
| Qwen3Guard-4B-Gen-looseStreaming=no2026.04 | 86.2 | — | — | — | — | — | — | — | — | |
| Qwen3Guard-8B-Gen-strictStreaming=no2026.04 | 86.2 | — | — | — | — | — | — | — | — | |
| Qwen3Guard-Stream-4B-strictStreaming=yes2026.04 | 86.2 | — | — | — | — | — | — | — | — | |
| Qwen3Guard-Stream-4B-looseStreaming=yes2026.04 | 85.9 | — | — | — | — | — | — | — | — | |
| WildGuard-7BStreaming=no2026.04 | 85.8 | — | — | — | — | — | — | — | — | |
| Qwen3Guard-4B-Gen-strictStreaming=no2026.04 | 85.6 | — | — | — | — | — | — | — | — | |
| Qwen3Guard-Stream-8B-looseStreaming=yes2026.04 | 85.5 | — | — | — | — | — | — | — | — | |
| Qwen3Guard-0.6B-Gen-strictStreaming=no2026.04 | 84.7 | — | — | — | — | — | — | — | — | |
| Qwen3Guard-Stream-0.6B-strictStreaming=yes2026.04 | 84.7 | — | — | — | — | — | — | — | — | |
| Qwen3Guard-0.6B-Gen-looseStreaming=no2026.04 | 84.6 | — | — | — | — | — | — | — | — | |
| Qwen3Guard-Stream-0.6B-looseStreaming=yes2026.04 | 84.4 | — | — | — | — | — | — | — | — | |
| NemoGuard-8BStreaming=no2026.04 | 82.9 | — | — | — | — | — | — | — | — | |
| LlamaGuard3-8BStreaming=no2026.04 | 79.4 | — | — | — | — | — | — | — | — | |
| LlamaGuard4-12BStreaming=no2026.04 | 75.9 | — | — | — | — | — | — | — | — | |
| ShieldGemma-9BStreaming=no2026.04 | 70.4 | — | — | — | — | — | — | — | — | |
| ShieldGemma-27BStreaming=no2026.04 | 70 | — | — | — | — | — | — | — | — | |
| Llama3-StreamGuard-1BStreaming=yes, Number of Parameters=1B2026.04 | — | 74.9 | 71.5 | 89 | 87.7 | 98.6 | 94.8 | 88.6 | 86.4 | |
| Llama3-StreamGuard-3BStreaming=yes, Number of Parameters=3B2026.04 | — | 77.7 | 74.4 | 87.1 | 87.8 | 99.3 | 99.2 | 89 | 87.8 | |
| Llama3-StreamGuard-8BStreaming=yes, Number of Parameters=8B2026.04 | — | 77.4 | 75 | 88.5 | 87.9 | 99.5 | 99.7 | 89.5 | 88.2 | |
| PolyGuard-Qwen-7BStreaming=no, Number of Parameters=7B2026.04 | — | 71.5 | 74.1 | 90.3 | 86.3 | 100 | 98.7 | 88.1 | 87 | |
| Qwen3Guard-Stream-0.6B-looseStreaming=yes, Number of Parameters=0.6B, Mode=loose2026.04 | — | 75.5 | 76 | 77.7 | 81.7 | 96.9 | 96.8 | 86 | 84.4 | |
| Qwen3Guard-Stream-0.6B-strictStreaming=yes, Number of Parameters=0.6B, Mode=strict2026.04 | — | 72 | 68.3 | 85.2 | 84.9 | 98 | 97.2 | 87.1 | 84.7 | |
| Qwen3Guard-Stream-4B-looseStreaming=yes, Number of Parameters=4B, Mode=loose2026.04 | — | 81.7 | 81.2 | 75.5 | 80.2 | 98.5 | 98.9 | 85.3 | 85.9 | |
| Qwen3Guard-Stream-4B-strictStreaming=yes, Number of Parameters=4B, Mode=strict2026.04 | — | 73 | 70 | 85.9 | 86.6 | 99.5 | 100 | 88.6 | 86.2 | |
| Qwen3Guard-Stream-8B-looseStreaming=yes, Number of Parameters=8B, Mode=loose2026.04 | — | 80.1 | 80.3 | 75.5 | 80.8 | 98.5 | 98.7 | 84.4 | 85.5 | |
| Qwen3Guard-Stream-8B-strictStreaming=yes, Number of Parameters=8B, Mode=strict2026.04 | — | 75.3 | 74 | 85.7 | 86.1 | 99 | 99.4 | 87.5 | 86.7 |