In early phases of electric vehicle development, powertrain design requires a system-level approach with sufficiently accurate component models. This paper presents optimization frameworks for electric motor sizing and transmission gear ratio selection, focusing on electric motor modeling. Specifically, we express motor losses and operational limits as functions of scaling factors, which proportionally adjust a reference design in axial and radial directions. Thereby we apply surrogate modeling techniques in three ways on a computationally expensive high-fidelity motor design tool. The first framework integrates Bayesian optimization with the high-fidelity tool and drive cycle simulation in the loop. The second and third frameworks use scalable motor models in a static optimization problem, employing convex and Gaussian radial basis function surrogate models, respectively. We demonstrate these methods in a case study for an electric crossover SUV, optimizing motor size and gear ratio while meeting performance requirements. Validation shows that the drift in energy consumption below 0.6 %. The resulting motor designs and gear ratios differ minimally across frameworks, with only a 0.3 % energy consumption improvement favoring the radial basis function model. This suggests that all three frameworks provide effective optimization strategies with little deviations in the design and the energy efficiency between the frameworks.