Osteoporosis is a “silent epidemic” of weakening bones, often leading to fracture. To predict fracture, the finite element method based on micro-computed tomography can be used, modelling bone behaviour under load. Elasto-plastic finite element modelling is able to directly calculate bone strength, but until recently it has been computationally expensive. Advancements in GPU technology have exponentially increased the computational power available in consumer-grade hardware and enabled the implementation of a custom elasto-plastic finite element solver for large-scale models. Both analytical and mechanical validation of the elasto-plastic solver was performed. Elasto-plastic finite element models based on tomography data were able to predict failure load with a correlation coefficient of 0.81. These new developments in elasto-plastic finite element modelling have enabled the direct estimation of patient-specific yield properties on consumer-grade hardware.