This shows powerful abilities in dealing with full process technology but leaves room for improvement in diff-like duties. None of the GPT-4o or Claude 3.5 Sonnets could response this easy problem properly. Only o1 was capable of finding the correct response with no guidance. Let’s see how Deepseek performs. Introduced https://x.com/kidtsang/status/1884008035535782292