No verify button when setting custom OpenAI API key in Settings → Models

[SAMYNHN]

Where does the bug appear (feature/product)?

Cursor IDE

Describe the Bug

I want to use my own OpenAI API key and model, but it doesn’t work.
When I put my key in Settings → Models → API Keys and set
https://api.openai.com/v1/ as the base URL, nothing happens — no verify, no error.

My API key is valid (it works fine with curl), but Cursor doesn’t connect or confirm it.

Steps to Reproduce

just install the latest version cursor

Screenshots / Screen Recordings

Operating System

MacOS

Current Cursor Version (Menu → About Cursor → Copy)

Version: 2.0.52
VSCode Version: 1.99.3
Commit: 2125c48207a2a9aa55bce3d0af552912c84175d0
Date: 2025-11-03T17:32:51.928Z
Electron: 37.7.0
Chromium: 138.0.7204.251
Node.js: 22.20.0
V8: 13.8.258.32-electron.0
OS: Darwin arm64 25.0.0

Does this stop you from using Cursor

Yes - Cursor is unusable

[deanrie]

Hey, thanks for the report. A couple of things to try:

You don’t need to override the base URL for official OpenAI API keys - Cursor uses https://api.openai.com/v1 by default. Try removing the base URL override completely and just enter your API key.

The verify button was intentionally removed in recent versions - you just need to enter your key and it should work automatically.

If you do need the base URL override for some reason, try without the trailing slash: https://api.openai.com/v1 (not https://api.openai.com/v1/)

Could you try entering just your OpenAI API key without any base URL override and let me know if it works?

[SAMYNHN]

Yes, I found that it works for some models like gpt-4-1,

but through my own OpenAI API key I should be able to access various models like gpt-5-pro, gpt-5, and gpt5-mini

However, it seems that these are being overridden or replaced by Cursor’s built-in models, and some of them didn’t show up either.

[deanrie]

Hey, thanks for the extra details. A couple of clarifications:

• Does your OpenAI account have API access to GPT-5 models? Check: https://platform.openai.com/docs/models
• Which models do you see in Cursor’s model picker after adding your API key?

This will help us tell if it’s a custom API key limitation or a configuration issue.

[SAMYNHN]

Yeah, I’m sure that my API key can access GPT-5 and other models — I can use them through Python scripts without any issues, and there’s no difference after adding my API key (I’ve already refreshed it and checked “view all models”)

NOTE: I also found that I can only use my own API’s gpt-t-nano and gpt-5-mini and gpt-4.1 models.

[SAMYNHN]

@deanrie Just following up in case you missed my previous message. May I ask if this issue can be resolved?

[AlbirrKarim]

i have enter my api key and it keep says “You’ve hit your usage limit Get Cursor Pro for more Agent usage, unlimited Tab, and more.“

[SAMYNHN]

I figured out the reason

Cursor blocks all GPT-5 and above models for free plan user, regardless of whether you use an API key or not.

@deanrie Can you help us fix this bug?

[deanrie]

Thanks for the clarification. Unfortunately, this isn’t a bug. To use your own API keys, you need an active subscription (Pro or higher). The free plan limits which models you can access, even when using your API keys.

[AlbirrKarim]

Thanks for the information, i think cursor should give some UI feedback about that, so user not confused.

[deanrie]

Thanks for the feedback. I agree that UI feedback would be helpful. The team should show a message like “This feature is available for Pro subscription only” instead of silently blocking it.