You will get rate limit errors on all the models if you push hard enough. I always wrap the API with some exponential backoff retries to work around this.