WebGen-Bench: Evaluating LLMs on Generating Interactive and Functional Websites from Scratch

TL;DR

WebGen-Bench is a comprehensive benchmark for evaluating large language models' ability to generate complex, multi-file websites from scratch, using diverse instructions and rigorous testing procedures.

Contribution

The paper introduces WebGen-Bench, a new benchmark with a large instruction set and testing framework to evaluate LLMs in website generation tasks.

Findings

The best model achieves only 27.8% accuracy, indicating the task's difficulty.

Training on WebGen-Instruct improves accuracy to 38.2%.

The benchmark reveals significant challenges in LLM-based website generation.

Abstract

LLM-based agents have demonstrated great potential in generating and managing code within complex codebases. In this paper, we introduce WebGen-Bench, a novel benchmark designed to measure an LLM-based agent's ability to create multi-file website codebases from scratch. It contains diverse instructions for website generation, created through the combined efforts of human annotators and GPT-4o. These instructions span three major categories and thirteen minor categories, encompassing nearly all important types of web applications. To assess the quality of the generated websites, we use GPT-4o to generate test cases targeting each functionality described in the instructions, and then manually filter, adjust, and organize them to ensure accuracy, resulting in 647 test cases. Each test case specifies an operation to be performed on the website and the expected result after the operation. To automate testing and improve reproducibility, we employ a powerful web-navigation agent to execute tests on the generated websites and determine whether the observed responses align with the expected results. We evaluate three high-performance code-agent frameworks, Bolt.diy, OpenHands, and Aider, using multiple proprietary and open-source LLMs as engines. The best-performing combination, Bolt.diy powered by DeepSeek-R1, achieves only 27.8\% accuracy on the test cases, highlighting the challenging nature of our benchmark. Additionally, we construct WebGen-Instruct, a training set consisting of 6,667 website-generation instructions. Training Qwen2.5-Coder-32B-Instruct on Bolt.diy trajectories generated from a subset of this training set achieves an accuracy of 38.2\%, surpassing the performance of the best proprietary model.