Paper Proposes Measuring LLM Creativity via Long-Tail Performance

A joint paper from University of Chicago, Stanford and Microsoft analyzes 26 open-source models across 15 creativity benchmarks, arguing that creativity tests essentially measure long-tail performance, unifying fragmented LLM creativity evaluation.

2026-10-14 ~ 2026-10-15 · 2 related posts

• LLM creativity lives in the long tail: UChicago study spans 15 benchmarks, 26 open models — Uchicago · 2026-10-14
• What creativity leaderboards actually measure: 15 benchmarks point to long-tail performance — shangbinfeng · 2026-10-15