ITBench-AA: Frontier Models Score Below 50% on the First Benchmark for Agentic Enterprise IT Tasks — by Artificial Analysis and IBM

In three linesITBench-AA, a new benchmark from Artificial Analysis and IBM, evaluates frontier models on agentic enterprise IT tasks. Top models (Claude, GPT-4, Gemini) score below 50%, exposing significant gaps in automating complex IT workflows.

Summary generated by Claude — human-verified