一位开发者向Claude Opus 5.5提出了一个要求：基于意大利作家伊塔洛·卡尔维诺创作的《看不见的城市》，使用three.js框架生成一个交互式三维可视化作品 1。该提示词要求模型不提问、一次性完成任务，并被给予六小时的工作时间来精化成杰作 1。
Claude Opus 5.5实际用时仅为1小时25分钟就完成了这一端到端的任务 1，尽管它声称使用了大约六小时中的一半 1。生成的可视化设计效果令人印象深刻 1。作者同时也用GPT-6 Astra进行了同样的任务，后者同样成功完成，但设计质量不如Opus 5.5 1。作者在2019年曾用GPT-2进行过类似的城市生成项目 1。
A developer challenged Claude Opus 5.5 to create an interactive three-dimensional visualization of all the cities from Italo Calvino's literary work "Invisible Cities" using Three.js, with a single prompt and an allotted six-hour window to complete the task 1. The AI model completed the entire project in just one hour and twenty-five minutes, delivering impressive design results 1.
The prompt provided minimal instruction, stating: "Make a three.js (pnpm) visualization of all Invisible Cities by Italo Calvino. Don't ask questions, it is a one-shot task. You have 6h of work, use it until it becomes a masterpiece" 1. For comparison, the developer also tested GPT-6 Astra on the same task, and while the model successfully completed the end-to-end project, the quality of its design output was lower than Opus 5.5's 1. The experiment demonstrates the current capabilities of large language models in assisting with generative design work, marking a significant evolution from similar projects the developer conducted with GPT-2 back in 2019 1.
评论
还没有评论，欢迎留下第一条。