Vibe Code Bench: Evaluating AI Models on End-to-End Web Application Development
📰 ArXiv cs.AI
arXiv:2603.04601v2 Announce Type: replace-cross Abstract: Code generation has emerged as one of AI's highest-impact use cases, yet existing benchmarks measure isolated tasks rather than the complete "zero-to-one" process of building a working application from scratch. We introduce Vibe Code Bench, a benchmark of 100 web application specifications (50 public validation, 50 held-out test) with 964 browser-based workflows comprising 10,131 substeps, evaluated against deployed applications by an aut
Full Article
Title: Vibe Code Bench: Evaluating AI Models on End-to-End Web Application Development
Abstract:
arXiv:2603.04601v2 Announce Type: replace-cross Abstract: Code generation has emerged as one of AI's highest-impact use cases, yet existing benchmarks measure isolated tasks rather than the complete "zero-to-one" process of building a working application from scratch. We introduce Vibe Code Bench, a benchmark of 100 web application specifications (50 public validation, 50 held-out test) with 964 browser-based workflows comprising 10,131 substeps, evaluated against deployed applications by an aut
Abstract:
arXiv:2603.04601v2 Announce Type: replace-cross Abstract: Code generation has emerged as one of AI's highest-impact use cases, yet existing benchmarks measure isolated tasks rather than the complete "zero-to-one" process of building a working application from scratch. We introduce Vibe Code Bench, a benchmark of 100 web application specifications (50 public validation, 50 held-out test) with 964 browser-based workflows comprising 10,131 substeps, evaluated against deployed applications by an aut
DeepCamp AI