Rubber-like polymers are valued for their unique mechanical properties, yet predicting their fracture remains a challenge due to the complex microscale dynamics of polymer networks. This difficulty is amplified in natural rubbers, where strain-induced crystallization significantly enhances toughness. This work introduces a comprehensive multiscale model to predict fracture in both strain-crystallizing and non-crystallizing rubbers. At the microscale, the model employs non-Gaussian statistics to capture entropic chain behavior while accounting for molecular bond distortions and crystallite evolution. These effects are bridged to the macroscale via a damage-adapted, non-affine microsphere model. A macroscopic phase-field approach is then used to simulate fracture, assuming that damage arises from chain failure and is hindered by local crystallinity. Comparisons with experimental data validate the model’s accuracy and its ability to capture the toughening mechanisms of diverse rubber-like materials.