Inference Intermediate Also known as: Abstraction and Reasoning Corpus

ARC-AGI

/ark-ay-jee-eye/

A benchmark of visual grid puzzles created by François Chollet to measure abstract reasoning on never-seen patterns, unsolvable by memorization.

ShareLinkedIn X

In practice

Designed to be easy for humans (over 80%) but hard for LLMs. In 2024 OpenAI's o3 hit historic results, reopening the debate on what AGI really means. A one-million-dollar prize is attached.

Related terms

Reasoning model Frontier model GPQA

Seen in the wild

2 entries mentioning it

February 19, 2026

Gemini 3.1 Pro: Google's first '0.1' bump and the ARC-AGI-2 leap

High
December 20, 2024

OpenAI o3: the model that beats ARC-AGI and redefines 'reasoning'

Landmark

← All terms