Purpose <p>Synchrotron-based X-ray computed tomography (CT) has become an essential tool for non-destructive, high-resolution 3D characterization in multiple scientific areas. However, the increasing scale of CT data demands efficient and user-friendly reconstruction software capable of handling large datasets. To address this need and considering the requirements of experiments, we developed MOCUPY—a GPU-accelerated reconstruction software, which is specifically designed for synchrotron CT data processing.</p> Methods <p>MOCUPY is composed of five functional modules with a PyQt5-based graphical interface. MOCUPY provides automatic and manual shift alignment modules to correct stage jitters for high-resolution reconstruction, multiple algorithms for phase retrieval of propagation-based phase-contrast images and reconstruction algorithms. The core algorithms of MOCUPY are implemented in a self-developed GPU-accelerated library named MOCU, which supports asynchronous execution, enabling simultaneous operations of GPU computation, file I/O, and memory transfer. The specialized pipeline avoids intermediate disk storage, and a large-data framework ensures efficient handling of TB-scale datasets.</p> Conclusion <p>MOCUPY achieves minute-scale reconstruction of large-scale CT data, satisfying the requirement for fast image processing of big-data CT and providing users with a high-performance solution.</p>