Other leading LLMs do answer the prompt correctly. This is just a meaningless exercise in kicking sand in OpenAI's face. (Well-deserved sand, admittedly.)