GPT models not working when quota reached but extra usage is approved

### Short summary

GPT models not working when quota reached but extra usage is approved

### Affected version or release

v.0.2.5

### Installation context

Local

### What happened?

It seems there is a problem that when the quota of premium requests is already reached, but the organization has extra usage approved, then any premium GPT model is getting a go away connection timeout from the server. BUT when starting the bundled GitHub Copilot CLI directly everything is working fine in that. So for me it seems a bug then inside the APP version itself. Btw all other models from different vendors are working perfectly fine in the app.

### Steps to reproduce

0. Premium Request Quota reached and github copilot app opened.
1. Choose any premium GPT model like GPT-5.4 and any effort level like medium
2. Choose plan or autopilot
3. give it any task to fullfill like creating a hello world app

### Expected behavior

GPT models can be used in the Github Copilot app despite premium quota reached and when extra usage is approved.

### Additional context

_No response_