The Frontier Design Benchmark is live.

Read on XLinkedIn

Dataset

Gallery

AccessBy request

The dataset isn’t public. For research or model evaluation, email us.

Dataset content

The dataset contains:

  • Real-world, project-based design tasks
  • Human golden design outputs
  • Model rollouts
  • Expert pairwise comparison preferences, with rationales (optional)
  • Design rubrics and principles
  • AI slop pattern list
Sign inGet in touch