In my Cursor Settings under Model Names, I swear I used to have the o1 family listed. But now, they’re gone.
“Add model” just opens a text input, typing “gpt-o1” into it does nothing.
I just created an OpenAI API key and verified it, just to be able to use o1 within Cursor. Is that no longer possible?
Hmm, how strange. Toggling the OpenAPI API Key off all of a sudden reveals a lot more models, among them o1. Which kinda defeats the point of using the key!
Is it the case that the key might not provide access to o1?
Possibly, I know there are some requirements. Trying adding o1 as a custom model and see what happens.
This might happen if you haven’t opened the settings page in a while and have only been using your API key. Models don’t update when the API is enabled. However, if you enable the API key now, this list should be there.
Adding “o1” manually let’s me at least choose it in the Chat. But it results in:
{"error":{"message":"The model `o1` does not exist or you do not have access to it.","type":"invalid_request_error","param":null,"code":"model_not_found"}}
Turns out this is apparently an OpenAI restriction: Why can't I access model o1 with my API key? - API - OpenAI Developer Forum
Which is a real shame, because on my project, o1 performs so much better than Claude etc…
o1-mini works though.
Furthermore, after enabling the OpenAI API key, it seems it gets used for requests to Claude as well, which is unexpected:
{"type":"error","error":{"type":"authentication_error","message":"invalid x-api-key"}}