Autonomous agents may generate millions of lines of code, but shipping software is another matter Opinion AI-integrated ...

A hands-on comparison shows how Cursor, Windsurf, and Visual Studio Code approach text-to-website generation differently once ...

VS Code forks like Cursor, Windsurf, and Google Antigravity may share a common foundation, but hands-on testing shows they ...

The unusual experiment, which was shared by Truell on X (formerly Twitter), involved the AI agents running uninterrupted for ...

整理 | Tina现在，大模型可以独立写完整整一个浏览器了？Cursor CEO Michael Truell 最近分享了一项颇为吸睛的实验：他们用 GPT-5.2 让系统连续不间断运行一周，从零构建出一个“可用”的 Web ...

知名 AI 编程辅助工具 Cursor 近日分享了一项内部测试发现:在处理超长程、自主性的编程任务时，OpenAI 最新的 GPT-5.2 模型表现出比 Anthropic 的 Claude Opus4.5 更高的可靠性。为了验证模型能力，Cursor 团队尝试从零开始构建一个功能完备的 Web 浏览器 ，涵盖了 HTML 解析、CSS 布局及自定义 JavaScript 虚拟机等复杂底层架构。

Cursor CEO says AI built a web browser in one week, but experts and users remain cautious about real-world usability.

It’s barely been a year since the term ‘vibe coding‘ was coined, but people are vibe coding entire browsers now. In a ...

An AI experiment used GPT-5.2 to build a 3M-line web browser in a week, revealing how far AI coding has come and sparking ...

Stability AI 前 CEO Emad Mostaque 则估算，构建这个浏览器可能消耗了约 30 亿个 Token。但随着 Token 成本持续下降，软件开发的边际成本正在趋近于零。

后来出现了所谓的Agentic编程——Claude Code、Cursor Agent、Windsurf等工具让AI可以自主执行多步任务，读取文件、运行命令、修复错误。 直到GPT-5.2。 Cursor团队让GPT-5.2持续运行了 整整一周 ，不是断断续续，而是 连续工作 。