ARC-AGI-3: A New Challenge for Frontier Agentic Intelligence
This provides a new challenge for assessing adaptive efficiency in AI agents, though it is incremental as it builds on prior ARC-AGI benchmarks.
The authors introduced ARC-AGI-3, an interactive benchmark for evaluating agentic intelligence in abstract, turn-based environments, where humans solve 100% of tasks while frontier AI systems score below 1% as of March 2026.
We introduce ARC-AGI-3, an interactive benchmark for studying agentic intelligence through novel, abstract, turn-based environments in which agents must explore, infer goals, build internal models of environment dynamics, and plan effective action sequences without explicit instructions. Like its predecessors ARC-AGI-1 and 2, ARC-AGI-3 focuses entirely on evaluating fluid adaptive efficiency on novel tasks, while avoiding language and external knowledge. ARC-AGI-3 environments only leverage Core Knowledge priors and are difficulty-calibrated via extensive testing with human test-takers. Our testing shows humans can solve 100% of the environments, in contrast to frontier AI systems which, as of March 2026, score below 1%. In this paper, we present the benchmark design, its efficiency-based scoring framework grounded in human action baselines, and the methodology used to construct, validate, and calibrate the environments.