Liuzhou Xingyuan Network Technology Co., Ltd. experimented with using AI models to build a corporate website, finding that relying solely on Gemini 3 Pro was insufficient for establishing the basic framework. They switched to Claude Opus 4.5 to complete the main development work, with Gemini 3 Pro later making minor modifications. The article reveals that the code quality generated by Claude Opus 4.5 was impressive, while Gemini 3 Pro’s code accounted for only about 20%. This practical case highlights the current differences in coding capabilities among AI models, offering valuable insights for developers engaged in AI-assisted development. By comparing the performance of two major AI models in a real-world project, the article provides a valuable reference for readers interested in AI programming technology, showcasing the current state and potential of AI in the software development field.
