Interesting discussion! It’s always helpful to see real-world experiences with large node counts instead of only theoretical limits. Performance clearly depends on model complexity, available RAM, solver settings, and hardware configuration. Thanks …