The ARC benchmark by Francois Chollet is designed to evaluate general intelligence in AI models through pattern-recognition and reasoning tasks.
