Qwen3Guard is a multilingual guardrail model series developed by the Qwen team at Alibaba Cloud.

A demo to show how the Guard model works

Implementation examples and usage demonstrations for Qwen3Guard-Gen-0.6B safety moderation model

No description available

4Billion Parameter Guardrail Model : Qwen3Guard-Gen-4B

Test for float16 quant deployment of Qwen3Guard using vLLM

Qwen3Guard-Gen-8B Model for Deployment

🛡️ Protect your AI models with Qwen3Guard, a powerful tool for ensuring security and reliability in your machine learning applications.

No description available

No description available

Testing gpt-oss:20b and qwen3guard to classify self harm edge cases

Evaluates roleplay jailbreak vulnerability across 6 open-source LLMs using persona-style AdvBench variants. LlamaGuard-3 as judge, Qwen3Guard as response-level defense. Shows alignment drives robustness more than model size.