When benchmarks saturate, what comes next? Meta’s GIM pushes AI evaluation toward integrated reasoning

Meta introduces GIM, a benchmark for integrated reasoning in AI evaluation.