The aim of the study is to predict compressive strength and the failure mechanism of concrete cubes considering the Tresca criterion for high-performance concrete reinforced with glass fibers and minibar basalt. The method of research includes experimental compression tests on 100 mm cubes at curing ages of 7, 14 and 28 days soft computing techniques such as linear regression ridge regression and polynomial regression implemented in Python using scikit-learn artificial neural networks developed with Python and NumPy to visualize cracks and failure patterns as pixels as well as the incorporation of the Mohr-Coulomb and Tresca criteria to model the plastic zone and pyramidal failure shape. The results demonstrate that polynomial regression provides the highest predictive accuracy for compressive strength achieving R2 values greater than 0.99 for minibar basalt fiber-reinforced high-performance concrete and greater than 0.93 for glass fiber-reinforced high-performance concrete while the artificial neural networks successfully identify and predict crack locations failure degree and pyramidal failure shapes and overall model performance is validated by a coefficient of determination greater than 0.97 with mean absolute error greater than 2.1 and root mean square error greater than 16.99 confirming strong agreement between experimental and predicted values. Therefore, the research's novelty lies in its use of artificial intelligence by taking into account the TRESCA criteria to precisely predict the properties of high-performance concrete and to understand the impact of its input factors, streamlining the process and providing useful tools for the real-world application of the model's capabilities in the field of civil engineering.