ITBench-AA：在首个面向企业IT智能体任务的基准测试中，前沿模型得分低于 AI资讯 · 4 阅读 · 2026-05-30 · 自动采集 ITBench-AA: Frontier Models Score Below 50% on the First Benchmark for Agentic Enterprise IT Tasks — by Artificial Analysis and IBM