Frontend engineering tasks where an AI builds working web applications from detailed specifications. Includes task definitions, runnable app files, evaluator logs, grades, and a summary report.
Available on Request · Sample Available
WebGen / Vibe Coding
Frontend engineering tasks where an AI builds working web applications from detailed specifications.
Volume
0
Volume
Scalable Volume
Coverage
Coding
53.3%Claude Opus 4.8 · Pass rate
Targets
WebGen-Bench
Modality & Delivery
TextCode→CodeText
Formats
What You Receive
Coverage
Package Structure
<project_id>/One folder per webapp build task
- brief.mdThe natural-language app brief given to the model
app/Reference implementation
- package.jsonDependencies and scripts
- src/Components, pages, and styles
- tests/Playwright checks the built app must pass
- rubric.mdDesign and UX grading guidance
Differentiators
- Ships runnable application files together with evaluator logs, grades, and a summary report.
Record Schema
Results
| Model | Metric | Result |
|---|---|---|
| Claude Opus 4.8 | Pass rate |
Methodology & Quality
Methodology
SME involvement
Domain-expert curated with up to L2 review
Stressors
Translating detailed specifications into complete working frontend applications.
Design criteria
- Evaluate both functional behavior and rendered appearance.
Build assumptions
Quality approach
- Functional subtests and rendered-appearance scoring are retained with the task artifacts.
Known limitations
Customization & Scalability
Custom sets available. Additional volume in progress
Get Started
Request a Sample
Share your target domains and the OTS team will follow up with sample data and next steps.