ResearchBenchmarksJob scheduling with permanent disruptions on τ2-bench+ REALM-bench+ P13 v1 (test)Follow3Success RateRAC_M-0.120.691.52.31May 5, 2026Evaluation ResultsMethodMethodLinksSuccess RateTokens Used (k)Time (s)Completion ScoreRAC_MFramework=RAC_MFramework=RAC_M2026.053887033RACFramework=RACFramework=RAC2026.05376580SagaLLMFramework=SagaLLMFramework=SagaLLM2026.052802020LGFramework=LGFramework=LG2026.05003—LG(PE)Framework=LG(PE)Framework=LG(PE)2026.0506643—