Abstract Reasoning

The Toddler Test That Stumps Frontier AI

Preschoolers beat GPT-o1, GPT-4V, and LLaVA on simple visual analogy tasks. The gap reveals something foundational about how children — and machines — actually reason about structure.

Maren Solis·March 17, 2026