EnterpriseOps-Gym

ServiceNow's benchmark of stateful, multi-step enterprise-operations environments where actions can be irreversible and success requires policy adherence, not just task completion (backfilled from March 2026). Gained independent adoption in July 2026 when Artificial Analysis launched EnterpriseOps-Gym-AA, running frontier models on it as a standing leaderboard — the industry-adoption signal that qualifies it for tracking.