{
  "added_at": "2026-10-17",
  "arxiv_id": "2603.26648",
  "arxiv_version_reviewed": "v3",
  "authors": [
    "Zehai He",
    "Wenyi Hong",
    "Zhen Yang",
    "Ziyang Pan",
    "Mingdao Liu",
    "Xiaotao Gu",
    "Jie Tang"
  ],
  "bibtex_url": "https://awesomerea.org/papers/citations/paper-vision2web.bib",
  "citation": {
    "container_title": "Proceedings of the 43rd International Conference on Machine Learning",
    "pages": "41393-41415",
    "publisher": "PMLR",
    "series": "Proceedings of Machine Learning Research",
    "sources": [
      "https://proceedings.mlr.press/v306/he26n.html"
    ],
    "volume": "306"
  },
  "code_url": "https://github.com/zai-org/Vision2Web",
  "date_basis": "Publication date stated for PMLR volume 306; the July 6–11 conference dates are not used as the publication date.",
  "date_basis_zh": "采用 PMLR 第 306 卷标明的出版日期；不将 7 月 6—11 日的会议日期作为出版日期。",
  "date_precision": "day",
  "description_en": "Benchmarks website development from visual prototypes and written requirements across 100 static-page, 66 interactive-frontend and 27 full-stack tasks, combining workflow-based functional checks with model-judged visual comparison.",
  "description_zh": "基于视觉原型和文字需求评测网站开发，包含 100 项静态页面、66 项交互前端与 27 项全栈任务，将基于工作流的功能检查与模型进行的视觉比较结合。",
  "doi": null,
  "entry_url": "https://awesomerea.org/papers/paper-vision2web/",
  "entry_url_zh": "https://awesomerea.org/zh/papers/paper-vision2web/",
  "evidence_urls": [
    "https://proceedings.mlr.press/v306/he26n.html",
    "https://proceedings.mlr.press/v306/",
    "https://raw.githubusercontent.com/mlresearch/v306/main/assets/he26n/he26n.pdf",
    "https://arxiv.org/abs/2603.26648",
    "https://arxiv.org/html/2603.26648v3",
    "https://vision2web-bench.github.io/",
    "https://github.com/zai-org/Vision2Web/blob/d80a4a3d0ea7326541ad87314294269a633789e4/README.md",
    "https://github.com/zai-org/Vision2Web/blob/d80a4a3d0ea7326541ad87314294269a633789e4/setup.py",
    "https://github.com/zai-org/Vision2Web/blob/d80a4a3d0ea7326541ad87314294269a633789e4/vision2web/evaluation/prompts.py",
    "https://github.com/zai-org/Vision2Web/blob/d80a4a3d0ea7326541ad87314294269a633789e4/vision2web/evaluation/functional_tester.py",
    "https://huggingface.co/datasets/zai-org/Vision2Web/blob/8f03299d92b9bd852e93852d0c21e8a4848ab661/README.md"
  ],
  "first_arxiv_date": "2026-03-27",
  "id": "paper-vision2web",
  "json_url": "https://awesomerea.org/papers/paper-vision2web/record.json",
  "kind": "research",
  "markdown_url": "https://awesomerea.org/papers/paper-vision2web/index.md",
  "markdown_url_zh": "https://awesomerea.org/zh/papers/paper-vision2web/index.md",
  "metadata_notes": "The final PMLR paper has seven authors and pages 41393–41415; arXiv v3 is dated 2026-07-20. Current artifact evaluation differs from the paper configuration. Licensing signals conflict: README text says CC-BY-NC-SA-4.0, Hugging Face metadata says Apache-2.0, the project page says CC BY-SA 4.0, and setup.py lists MIT while the repository has no LICENSE file. No uniform artifact license is asserted.",
  "metadata_notes_zh": "正式 PMLR 论文包含七位作者，页码为 41393–41415；arXiv v3 日期为 2026-07-20。当前产物的评测配置与论文不同。许可信息存在冲突：README 正文声明 CC-BY-NC-SA-4.0，Hugging Face 元数据标为 Apache-2.0，项目页声明 CC BY-SA 4.0，setup.py 列出 MIT，但仓库没有 LICENSE 文件。因此不宣称产物适用统一许可。",
  "pdf_url": "https://raw.githubusercontent.com/mlresearch/v306/main/assets/he26n/he26n.pdf",
  "project_url": "https://vision2web-bench.github.io/",
  "publication_date": "2026-09-29",
  "publication_status": "conference-paper",
  "record_type": "paper",
  "runtime_tested": false,
  "scope_limitations_en": "The full-stack tier is specification-driven, not reconstruction solely by exploring a live website. Workflows test state management and CRUD outcomes, but finite agent judgments do not establish exact backend or durable-database equivalence. The current public evaluator maintains a browser session within each workflow and permits local business-API checks; this is not a universal persistence proof.",
  "scope_limitations_zh": "全栈层级由明确规格驱动，并非仅通过探索运行中的网站进行重建。工作流测试状态管理与增删改查结果，但有限的智能体判断不能证明后端或数据库持久性的精确等价。当前公开评测器在每个工作流内保持浏览器会话，并允许检查本地业务 API；这不构成通用的持久化证明。",
  "targets": [
    "website"
  ],
  "title": "Vision2Web: A Hierarchical Benchmark for Visual Website Development with Agent Verification",
  "title_zh": "Vision2Web：采用智能体验证的分层视觉网站开发基准",
  "url": "https://proceedings.mlr.press/v306/he26n.html",
  "venue": "ICML 2026",
  "verification": "source-reviewed",
  "verified_at": "2026-10-17",
  "why_fits_en": "A useful comparison for evaluating interactive and stateful website reconstruction, while separating written-specification development from behavior-only inference.",
  "why_fits_zh": "可用于比较交互式、有状态网站重建的评测方法，同时明确区分根据文字规格开发与仅凭行为推断。",
  "year": 2026
}