QwenLM/Qwen ? reverse-engineered prompt

Build me a simple local app for chatting with Qwen, with both a web demo and a command line demo so I can try the model easily on my own machine.

I want it to load one of the Qwen chat models, send prompts, stream the replies if possible, and make it feel like a normal chatbot for asking questions, summarizing text, translating, and basic coding help. Please also include a small OpenAI style API server so other tools can talk to it the same way they would to ChatGPT.

If it makes sense, add support for running a quantized model too so it uses less memory, and make the setup instructions easy to follow. Keep the interface clean and simple, and if you need to check current model or dependency docs online, go ahead.