Lithium-ion batteries (LIB) are synonymous with the modern age of electrification, yet advances in battery design, manufacturing, and chemistry are still urgently needed. Mathematical modelling plays an important role in understanding LIB performance and can provide physics informed design directions, optimisation and explain outcomes. We present an exploration and detailed comparison of the commonly used homogenised Doyle-Fuller Newman (DFN) model and the high fidelity X-ray computed tomography (CT) based microstructural model for LIBs. We provide insights into the relative benefits of each model and highlight why they are important to battery technology development. Alongside experiments, we use the models to explore and compare two common cathode chemistries, lithium nickel manganese cobalt oxide, Li[Ni0.6Co0.2Mn0.2]O2 (NMC622), and lithium iron phosphate, LiFePO4 (LFP), and investigate the influence of electrode thickness and discharge current density. The DFN and CT image-based models show good alignment for averaged LIB metrics, such as the voltage response and active material utilisation, demonstrating that homogenised, computationally inexpensive models are an essential basis for battery design and optimisation. The CT-based microstructural model provides further insight into localised particle and electrode dynamics, taking into account heterogeneities that are a source of battery degradation. Qualitatively, the models also compare well with experimental secondary ion mass spectrometry (SIMS) mapping of the Li concentration in the active particles across the electrode thickness.