Isn't the 5+ model version of openai is a reasoning one so that's why they reply slower then other models like Anthropic?