There is no parallel. If meta required you to be logged in to run local inference with llama, that might be similar. Your using openai's cloud so you're logged in.