模型会在输出的数字与汉字之间强行添加一个空格，导致文件读取失败 the model forcibly inserts a space between the numbers and Chinese characters in the output, causing the file to fail to read #2032
Description
What happened?
模型会在输出的数字和汉字之间添加一个空格，即使我要求它不加空格，他依然加了一个空格，始终不能去除。
经验证，此bug可导致读取含有“数字+汉字”名称组合的文件时出现严重错误，完全无法读取这样的文件，希望能尽快修复。
我使用的模型是该工具免费提供的内置的Qwen3.5-Plus模型，通过网页授权，但在网页版Qwen Chat中不会出现此问题。
The model adds a space between the numbers and Chinese characters in the output, even though I requested it not to add a space, it still adds a space, and it cannot be removed.
Upon verification, this bug can cause severe errors when reading files with the name combination of numbers + Chinese characters, making it impossible to read such files. I hope it can be fixed as soon as possible.
I am using the free built-in Qwen3.5-Plus model provided by this tool, which is authorized through the web page, but this issue does not occur in the web version of Qwen Chat.
以下为测试过程的图像：
What did you expect to happen?
应当是在需要时才在数字和汉字之间添加一个空格，而不是始终添加。
Client information
Qwen Code: 0.11.0 (a31cf75)
运行环境: Node.js v24.0.2 / npm 11.8.0
操作系统: win32 x64 (10.0.22631)
认证: qwen-oauth
模型: coder-model
会话 ID: 4f12e9e4-b5cc-402a-8d90-666835396171
沙箱: no sandbox
代理: no proxy
内存使用: 192.6 MB
Login information
No response
Anything else we need to know?
No response