Existing code-generation benchmarks score a single mapping from a complete prompt to a one-shot output. However, real web development is different. Users seldom write a full spec at the start; many requirements only become clear once they look at an intermediate result and react to it. We present Asuka-Bench, a benchmark that pairs underspecified user intent with multi-round refinement, grounded in browser-rendered behavior. Each task is resolved through a closed loop: a Code Agent generates a web project, a UI Agent executes test cases on the deployed site, and a User LLM turns evaluation outcomes into natural-language feedback for the next round. The benchmark comprises 50 web tasks with 784 evaluation criteria and 2402 expected outcomes. We benchmark 8 LLMs across 2 agent frameworks. The results separate models clearly: weighted Task Pass Rate varies by 38 percentage points and models also differ substantially in their ability to repair from feedback. Asuka-Bench is also far from saturated: even the strongest model completes only 52% of projects after three rounds.

翻译：现有代码生成基准测试仅评估从完整提示到单次输出的单一映射。然而，实际Web开发截然不同：用户很少在初始阶段编写完整规范，许多需求只有在查看中间结果并做出反馈后才逐渐明确。为此，我们提出Asuka-Bench——一个以浏览器渲染行为为基础，将欠指定用户意图与多轮精修相结合的基准测试。每个任务通过闭环流程完成：代码智能体生成Web项目，UI智能体在部署站点上执行测试用例，用户大语言模型将评估结果转化为面向下一轮的自然语言反馈。该基准测试包含50个Web任务、784项评估准则及2402个预期结果。我们在两个智能体框架下对8个大语言模型进行了评测，结果清晰区分了各模型性能：加权任务通过率差异达38个百分点，且模型从反馈中修复缺陷的能力也显著不同。Asuka-Bench远未达到饱和状态：即便最强模型在三轮迭代后也仅完成52%的项目。