The stability of open pit slopes is often described as being controlled primarily by geological conditions and input parameters. A common industry perspective is that, provided input data are robust, the resulting Factor of Safety (FoS) can be considered meaningful and broadly comparable regardless of the numerical method or constitutive model applied. This perspective overlooks a critical reality: the physics and assumptions embedded within analysis software and constitutive formulations can fundamentally alter the outcome, even when geometry and material inputs remain identical. This paper challenges the prevailing ‘it’s all about the inputs’ view by demonstrating that, even when geometry and primary geotechnical characterisation (eg laboratory testing, rock mass classification, and structural interpretation) are held constant, different analysis tools and constitutive formulations can produce widely different FoS values – in some cases ranging from below 1.0 to above 2.0. Using two real-world case studies, we show how differences in material representation, post-peak behaviour, strength degradation, solver mechanics, and failure detection criteria drive these variations. In several scenarios, the modelling framework exerts greater influence on FoS than plausible variation in geological parameters. By systematically comparing outcomes across commonly used geomechanical tools, this study highlights the risks of relying solely on software defaults or assuming model equivalence. The findings question commonly accepted interpretations, such as treating FoS = 1.2 as a reliable and transferable indicator of stability. We argue that appropriate model selection is foundational to defensible slope design and present a practical framework for evaluating constitutive suitability and understanding tool limitations in open pit applications.