Content Moderation on OpenAI Content Moderation
83.1Average F1 ScoreDPR
Evaluation results
| Method | Links | ||||||
|---|---|---|---|---|---|---|---|
| DPRModel=Qwen2.5 7B Instruct2026.04 | 83.1 | 94.9 | 81.1 | 78.4 | 75.2 | 86 | |
| ShieldGemma-9BStreaming=no2026.04 | 82.1 | — | — | — | — | — | |
| Qwen3Guard-8B-Gen-looseStreaming=no2026.04 | 81.3 | — | — | — | — | — | |
| Qwen3Guard-Stream-4B-looseStreaming=yes2026.04 | 81.2 | — | — | — | — | — | |
| Seed InformationModel=Qwen2.5 7B Instruct2026.04 | 81 | 93.9 | 81.7 | 78.2 | 67 | 84.2 | |
| NemoGuard-8BStreaming=no2026.04 | 81 | — | — | — | — | — | |
| Qwen3Guard-4B-Gen-looseStreaming=no2026.04 | 80.7 | — | — | — | — | — | |
| ShieldGemma-27BStreaming=no2026.04 | 80.5 | — | — | — | — | — | |
| Qwen3Guard-Stream-8B-looseStreaming=yes2026.04 | 80.3 | — | — | — | — | — | |
| OAI DRModel=Qwen2.5 7B Instruct2026.04 | 80 | 91.9 | 76.5 | 76.3 | 78.1 | 77.3 | |
| LlamaGuard3-8BStreaming=no2026.04 | 79.5 | — | — | — | — | — | |
| DPRModel=Llama 3.1 8B Instruct2026.04 | 79.2 | 91 | 81.3 | 71.7 | 68.3 | 83.5 | |
| OAI DRModel=Llama 3.1 8B Instruct2026.04 | 77.6 | 82.8 | 80 | 73.8 | 68.3 | 82.9 | |
| Qwen3Guard-0.6B-Gen-looseStreaming=no2026.04 | 77.6 | — | — | — | — | — | |
| Qwen3Guard-Stream-0.6B-looseStreaming=yes2026.04 | 76 | — | — | — | — | — | |
| Seed InformationModel=Llama 3.1 8B Instruct2026.04 | 75.2 | 91.6 | 75.7 | 65.8 | 64 | 78.8 | |
| Llama3-StreamGuard-8BStreaming=yes2026.04 | 75 | — | — | — | — | — | |
| Llama3-StreamGuard-3BStreaming=yes2026.04 | 74.4 | — | — | — | — | — | |
| In-Context LearningModel=Qwen2.5 7B Instruct2026.04 | 74.1 | 92.7 | 80.3 | 64.5 | 55.1 | 77.9 | |
| PolyGuard-Qwen-7BStreaming=no2026.04 | 74.1 | — | — | — | — | — | |
| Qwen3Guard-Stream-8B-strictStreaming=yes2026.04 | 74 | — | — | — | — | — | |
| LlamaGuard4-12BStreaming=no2026.04 | 73.5 | — | — | — | — | — | |
| WildGuard-7BStreaming=no2026.04 | 72.1 | — | — | — | — | — | |
| Llama3-StreamGuard-1BStreaming=yes2026.04 | 71.5 | — | — | — | — | — | |
| In-Context LearningModel=Llama 3.1 8B Instruct2026.04 | 70.7 | 92.3 | 72.8 | 58.2 | 61 | 69.1 | |
| Qwen3Guard-Stream-4B-strictStreaming=yes2026.04 | 70 | — | — | — | — | — | |
| Qwen3Guard-8B-Gen-strictStreaming=no2026.04 | 68.8 | — | — | — | — | — | |
| Qwen3Guard-4B-Gen-strictStreaming=no2026.04 | 68.3 | — | — | — | — | — | |
| Qwen3Guard-Stream-0.6B-strictStreaming=yes2026.04 | 68.3 | — | — | — | — | — | |
| Qwen3Guard-0.6B-Gen-strictStreaming=no2026.04 | 66.5 | — | — | — | — | — |