But you can easily make requests to GPT servers with an Execute Code card! In the request you can use whatever model you prefer, including GPT-4