I ran Microsoft Copilot against Meta AI, Code Llama, Gemini Advanced, and ChatGPT. It managed to get every test wrong. That's some kind of record or something.