Qwen/Qwen3Guard-Gen-8B

Qwen/Qwen3Guard-Gen-8B is an 8 billion parameter safety moderation model developed by Qwen, built upon the Qwen3 architecture. It is specifically designed for generative safety classification, framing it as an instruction-following task. This model excels at detailed risk assessment with a three-tiered severity classification (safe, controversial, unsafe) and offers robust multilingual support across 119 languages for both prompt and response moderation.