Inclusions trapped in diamonds provide pivotal information for investigating the Earth's interiorconditions at the time of diamond formation. The depth at which these inclusions areencapsulated is crucial, shedding light on the intricate history of diamond growth andsubsequent exhumation to the Earth’s surface. The entrapment pressures of these inclusions,obtained through elastic geothermobarometry methods (Kohn et al., 2023; Rustioni et al., 2015)reveal the pressures at which they were enclosed within the diamond. However, applying thismethod requires certain assumptions, notably assuming all volume changes post-entrapment tobe elastic. Viscous deformation or diamond cracking subsequent to growth can release inclusionstress, impacting measured pressures which ultimately can lead to underestimation of growthdepths(Angel et al., 2022; Rustioni et al., 2015). Due to the fast exhumation of diamonds tothe Earth’ surface, viscous relaxation is often assumed to be negligible. Therefore, we focusedon the modeling of brittle failures developed in diamonds around their inclusions, in order toexplore the conditions at which fractures may occur in diamonds and to evaluate the associatedstress release.We utilized Phase-Field Modeling to analyze fracture propagation and quantify pressure dropresulting from brittle fracture. Our implementation involved an ABAQUS UEL equipped withthe BFGS quasi-Newtonian monolithic algorithm, utilizing the history field irreversibilityapproach and AT2 damage model. Our study culminated in a comparative analysis betweenPhase-Field Modeling and cohesive zone modeling (CZM)-based discrete models (XFEM).Based on our numerical tests, we found that brittle fracture relaxation accounts for less than 9%of total elastic relaxation, indicating its limited significance. Additionally, we explored howinclusion size, shape, fracture strength, and toughness affect fracture initiation and propagation.This study pioneers phase-field modeling in mineral physics, to predict microscale fractureswithin a 3D multiaxial loading structure. Furthermore, our comparison with XFEM-CZM addsinsights into their applicability in inclusion-matrix problems, highlighting existingdiscrepancies for further discussion as outlined by Wu et al., 2020 .