OpenAI co-founder Kapathi proposed a new approach to AI benchmarking: providing the model with the opening text of The Lord of the Rings, requiring it to c...
技巧观点IT之家(RSS)
Today AI Intelligence Brief
OpenAI co-founder Kapathi proposed a new approach to AI benchmarking: providing the model with the
opening text of The Lord of the Rings, requiring it to create an interactive 3D world. He used 1 million tokens (costing about $10) to build scenes using three.js for Claude Opus 5. The model took about 2 hours to output approximately 5,500 lines of code, generating scenes such as Bilbo's farewell banquet, with a high level of completion.
Original Article Excerpt
IT之家:https://www.ithome.com/ 8 月 5 日消息,OpenAI 联合创始人安德烈 · 卡帕西(Andrej Karpathy)于 8 月 2 日在 X 平台发布推文,提出一项 AI 基准测试思路: 向模型提供《指环王》开篇首段文字,再要求模型创建对应的 3D 世界。
在测试 AI 模型时,此前业内习惯使用生成“鹈鹕骑车图”的 SVG 图片方式,用来评估大语言模型在空间推理、几何逻辑和复杂代码生成方面的真实水平。
不过卡帕西指出目前模型能力不断提升,等简单任务已不在话下,因此提出新的 AI 基准测试思路: 让模型根据文学文本构建可交互的 3D 场景,从而观察模型在代码生成、场景设计与多媒体制作方面的综合表现。
Karpathy 要求 Claude Opus 5 使用 three.js 创建 3D 世界,输入内容为《指环王》的首段文字。测试的最高资源额度为 100 万 tokens,成本约为 10 美元。在这一配置下,Claude Opus 5 耗时约 2 小时,输出约 5500 行代码。
生成作品以 3D 动画形式呈现《指环王》开场相关场景,其中包括人们参加比尔博告别宴会,以及充满财宝的洞穴。报道指出,该作品无法与人工耗费数小时制作的内容相比,但考虑到 AI 在约 2 小时内完成生成,其完成度仍较高。
Intelligence Assessment
Aioga 编辑摘要
卡帕西提出以文学文本生成可交互3D世界的测试思路,并让Claude Opus 5依据《指环王》开篇首段、使用three.js构建场景。测试额度为100万tokens,成本约10美元,约2小时产出约5500行代码。
The readable text on this page was extracted from the public source and organized with attribution, publication time and the original link. Copyright remains with the original author and publisher.
OpenAI co-founder Kapathi proposed a new approach to AI benchmarking: providing the model with the opening text of The Lord of the Rings, requiring it to c...
IT之家(RSS)2026-08-05T06:32:06.000Z
Scan to open this article
Aioga aggregates global AI updates and preserves source information for verification and citation.