← Back to Benchmark

Benchmark Dataset

Source data used for comparing Claude Opus 4.8 and Fable 5 across two prompting styles.

Contents

Add your dataset files to this folder. Both models are evaluated on the same inputs using Carving with Stone (rigid, schema-first prompts) and Carving with Clay (iterative, exploratory prompts).

  • Place CSV, JSON, images, or other source files here
  • Update this page with download links as you add files
  • See README.md for folder conventions