This doesn't seems to make sense? OpenAI client does not list llama 3 as one of the support model.