Bug: Default models hang at “Planning next step” after setting custom OpenAI API, while Auto mode and custom models work normally

Support Bug Reports
1

Where does the bug appear (feature/product)?

Cursor IDE

Describe the Bug

After set custom openai API and do some test, all default models do not response, no matter what models I choose, it shows planning the next step and did nothing. However, the auto mode can work correctly, and the custom model works fine. I tried to disable custom api and api key, remove the config(located at ~/Library/Application\ Supported/Cursor), reinstall cursor, nothing helps.

Steps to Reproduce

  1. add a custom api(use litellm proxy) and api key
  2. add a custom model and disable all other models
  3. test the custom model(it works fine now)
  4. enable some of the default models, try to do the test firstly, all models stuck at planning the next step

Expected Behavior

The default model I choose can work properly

Operating System

MacOS

Current Cursor Version (Menu → About Cursor → Copy)

Version: 2.2.43
VSCode Version: 1.105.1
Commit: 32cfbe848b35d9eb320980195985450f244b3030
Date: 2025-12-19T06:06:44.644Z
Electron: 37.7.0
Chromium: 138.0.7204.251
Node.js: 22.20.0
V8: 13.8.258.32-electron.0
OS: Darwin arm64 25.1.0

For AI issues: which model did you use?

Model: Opus 4.5, Gemini 3 Pro, GPT 5.2

For AI issues: add Request ID with privacy disabled

There are no request id

Does this stop you from using Cursor

Yes - Cursor is unusable

2

Hey, thanks for reporting this.

This looks like a conflict between your custom API setup and the default models. Since Auto mode and custom models work fine, there may be leftover settings affecting default model routing.

Could you please try:

  1. Go to Cursor Settings > Models and make sure the Custom OpenAI Base URL and API key fields are completely empty (not just disabled).
  2. Run network diagnostics: Cursor Settings > Network > Run Diagnostics and share the results.
  3. After removing the custom API settings, start a new chat (not just clear the context).
  4. If the issue persists, share the console output: Help > Toggle Developer Tools > Console tab (filter for errors).

Let me know what you find.

3

Ah, I toggled developer tools, it seems I have ran out of my usage (Err resource_exhausted), but strangely, it doesn’t pop up a warning window, I remember there should be one. And additional, before I added custom endpoint, my resources are all normal, but when I test the custom endpoint in ask mode, there was a warning told me that I have ran out of my limit.

I checked the dashboard and it doesn’t tell me how many free token left, does cursor change the subscription policy recently?

Anyway, thanks a lot for your help, it really helps.

image
image1932×1192 162 KB