Artificial Analysis has introduced a new agentic benchmark that evaluates large language models' ability to plan and execute tasks. Claude Fable and GLM 5.2 achieved top positions within their respective cohorts, demonstrating strong performance on this unsaturated benchmark.