Benchmark RadarAGC-Bench
AGC-Bench evaluates artificial general creativity across 78 datasets covering brainstorming, problem solving, STEM, narrative, figurative language, and humor. It uses an agentic harness and a public leaderboard with an open-weight judge model.
- Released
- 2026-07-01
- Readiness
- Paper only
- Primary field
- General AI
Why it matters
Creativity is a key aspect of intelligence often overlooked in AI evaluation. AGC-Bench provides a comprehensive, standardized measurement of AI creativity, revealing distinct strengths and weaknesses across models and a single creativity factor analogous to general intelligence.
Motivation
Creativity research has debated whether creativity is domain-specific (e.g., visual, writing, science), and if it is psychometrically separable from general intelligence.
Primary resources
Benchmark Radar records only publicly supported details and links back to primary sources for verification.