ARC-AGI-3

ARC-AGI-3 is an interactive reasoning benchmark that challenges AI agents to explore novel environments, acquire goals on the fly, build adaptable world models, and learn continuously. It measures agent performance by tracking skill acquisition efficiency, long-horizon planning, and experience-driven adaptation rather than only final answers. The benchmark provides replayable runs, a developer toolkit for agent integration, and an interactive UI for transparent evaluation. Participants test agents in environments designed to be solvable by humans but to expose gaps in current AI learning.

Access to a developer toolkit for agent integration, an interactive UI for testing, and replayable runs for inspection and evaluation of agent behaviour.

Integrate your agent using the ARC-AGI-3 toolkit and use the interactive UI to test and iterate the agent.