是否已有关于该错误的issue或讨论？ | Is there an existing issue / discussion for this?
该问题是否在FAQ中有解答？ | Is there an existing answer for this in FAQ?
当前行为 | Current Behavior
The platform is extremely slow and unstable. I am constantly getting HTTP 504 Gateway Timeout errors. When the model does respond, it cuts off mid-sentence or fails to complete simple instructions (like generating a list of 40 words). This is happening across all models (2.5, 3.5, and 3.6 Plus).
期望行为 | Expected Behavior
I expect a stable connection without server timeouts. The model should be able to finish its output completely and maintain a reasonable generation speed.
复现方法 | Steps To Reproduce
Go to chat.qwenlm.ai.
Ask a simple prompt (e.g., "Give me 40 words with 'h' in the middle").
The system either hangs (latency), triggers a 504 Error, or stops generating text halfway.
运行环境 | Environment
- OS: Windows 10
- Browser: Google Chrome
- Interface: Official Web Chat (https://qwen.ai)
- Python/Transformers/CUDA: N/A (Using Web Interface)
备注 | Anything else?
This issue is consistently happening for users in my region. It seems like a major infrastructure or server-side problem.
是否已有关于该错误的issue或讨论？ | Is there an existing issue / discussion for this?
该问题是否在FAQ中有解答？ | Is there an existing answer for this in FAQ?
当前行为 | Current Behavior
The platform is extremely slow and unstable. I am constantly getting HTTP 504 Gateway Timeout errors. When the model does respond, it cuts off mid-sentence or fails to complete simple instructions (like generating a list of 40 words). This is happening across all models (2.5, 3.5, and 3.6 Plus).
期望行为 | Expected Behavior
I expect a stable connection without server timeouts. The model should be able to finish its output completely and maintain a reasonable generation speed.
复现方法 | Steps To Reproduce
Go to chat.qwenlm.ai.
Ask a simple prompt (e.g., "Give me 40 words with 'h' in the middle").
The system either hangs (latency), triggers a 504 Error, or stops generating text halfway.
运行环境 | Environment
备注 | Anything else?
This issue is consistently happening for users in my region. It seems like a major infrastructure or server-side problem.