What changes if I enable my own API key for OpenAI models?

Does it affect autocomplete? Can I modify the system prompts? Basically, what changes when I enable an API key for use inside Cursor?

1 Like

Hi @RationallyPrime

If you don’t have the Pro plan, using your API key limits access to certain features like Cursor Tab, Composer, and Apply code. The other functions should work properly. What advantages do you see in using your key?

@deanrie , hi. If I have Pro plan and turn on my own API key from OpenAI, will autocomplete feature work with it? Or autocomplete works only with Cursor custom models?

Hi, if you have a Pro subscription and activate your API key, autocomplete will work. But yes, this is our own model. Without a subscription, even if you have a key, it won’t work.

The #1 advantage is not hitting rate limits. I was chatting with Gemini 1206 and ran into this:

Hit Google rate limit
We’ve hit a rate limit with Google. Please try again in a few moments.

I’m trying to understand what’s going on here…

  1. The Cursor IDE was using Gemini for the chat. Not its own custom model, right? That’s what the top red box in the screenshot indicates, gemini-exp-12016.
  2. Cursor’s backend proxies my chat prompt to Google, and hits an API rate limit. With so many users, totally understandable.
  3. I’m happy to provide my own Google AI Studio API key. Why would that disable anything else?

If the IDE was using gemini-exp-12016 for the chat with Cursor’s backend API key, can it use it with my own Google API key? Isn’t the magic in the prompt?