Qwen not outputting for relatively simple single file refactoring #597

I have a C# file with a little more then 1500 lines of code, that has 8 private members that have value assigned to it but never read. This is know as "IDE0052"

Using VSCode, and the chat feature, I selected the local Qwen-Coder model running locally on my M5 pro with 48gb of memory and told it:

Fix only the IDE0052 diagnostic in the current file.

Make the smallest possible edit.
Do not search the repository.
Do not run build or tests.
Do not inspect unrelated code.
Do not explain the warning before editing.

/no_think

ollama chugged over 25gb of memory, the GPU went 100% and for the last 45+ minutes it has been reading the 1500 lines file without doing anything...

the expected outcome is a simple deterministic listing of the private members followed by a 1 by 1 checking if its being used...

For reference, GPT-5.3-codex concludes this in less than 15 seconds.

What am I doing wrong?

Setup:
Hardware: Macbook Pro M5 with 48gb memory
Model: Qwen coder 30b
Context Size: 128kb
IDE: VS Code
Extension: Copilot
No additional software running, basically 40gb of available memory