I had the same issue and what worked for me was cancelling my current billing plan and then creating a new plan that uses prepaid billing instead of pay-as-you-go which is what my previously plan was. After that I was able to test a few different ways that I now have access to GPT4 via the API.
Once that was sorted, I went to Cursor and togged off the “Using key” under the OpenAI API setting, and then quit Cursor completely and relaunched it. After it reopened I went in and toggled the API back on, closed out of that settings page, reloaded the Cursor UI (search ‘reload window’ via the command palette), after which I could now select gpt-4 in the chat. I was able to confirm by asking what gpt model it’s using and it responded with GPT-4.
Seems like OpenAI changed their billing type from pay-as-you-go to prepaid billing, but might have only been enforcing it for new accounts. But since they then added the “successful payment of $1 or more” constraint to be able to access the GPT-4 API, it was now blocking accounts that had previously been able to use GPT-4 API just fine and didn’t provide any messaging about what was going on.
Hope that works for you like it did for me