Ultra-deep carbonate reservoirs were characterized by extreme temperatures and high closure stresses, where existing wellhead devices frequently encountered fracturing initiation challenges during stimulation operations. Accurate prediction of fracture pressure was recognized as crucial for developing effective pressure-reduction methodologies, though conventional prediction models were found insufficient due to inadequate consideration of thermal effects. To address these limitations, an integrated dynamic fracture pressure prediction model was established through coupled temperature-stress-seepage analysis, incorporating both thermal effects and fracturing fluid filtration mechanisms. Systematic investigations were conducted on the influences of formation properties, operational parameters, and perforation characteristics on fracture pressure in ultra-deep wells. The research findings demonstrate that after incorporating thermal effects, the model was demonstrated to accurately simulate downhole pressure and temperature variations during the hydraulic fracturing process in case well P103. Notably, a matching degree of 98.98% was achieved between predicted and measured fracture pressures, with predictive accuracy being improved by 11.66% compared to conventional models neglecting thermal considerations. The greater the elastic modulus of the formation, the higher the permeability, the greater the porosity, the lower the viscosity of the working fluid, the smaller the injection volume, the longer the perforation length, and the smaller the well inclination angle, all will lead to a decrease in the rupture pressure, while the influence of Poisson's ratio is relatively small. In addition, the optimal perforation orientation decreases as the azimuth of the wellbore increases. Based on this research, measures such as increasing permeability and porosity, using low-viscosity working fluid, reducing injection volume, extending perforation length and optimizing perforation orientation are proposed to achieve the goal of reducing rupture pressure. It is evident that the establishment of an integrated temperature-stress-seepage dynamic prediction model for rupture pressure is of great significance for formulating measures to reduce rupture pressure and successfully breaking the reservoir.