In magnetic device modeling, the crystallographic textures and microstructures of electrical steels critically influence their magnetic properties. Yet, these aspects are often neglected or inadequately represented in finite element (FE) models, leading to inaccuracies in predicting electrical steel performance, especially in complex electromagnetic environments. This paper presents a multi-scale modeling approach, integrating a macroscopic FE model of a magnetic device with a microstructural scalar permeability model that accounts for the crystallographic textures of electrical steels. This approach demonstrates proficiency in predicting nonlinear anisotropic permeability and capturing magnetic anisotropy of electrical steels within a rotational tester, especially when anisotropy is predominantly governed by crystallographic texture and the angles between the B and H vectors are minimal. Furthermore, the model effectively captures the magnetic anisotropy in complex magnetic fields and geometric configurations of a permanent magnet motor. A comparative analysis of this model is conducted against three other material models: an isotropic model, a two-axis model, and a general-vector model. These models are based on different approaches, ranging from a single non-linear BH curve to comprehensive vector BH curves covering all directions. Among them, the general-vector model, reliant on extensive 2D vector BH measurements, is the most accurate but requires substantial experimental data.