Pick a model, send a prompt, and copy the exact API call. (demo responses — add your API key for live output)