<p>Origami structures enable unique mechanical properties through folding motions. Due to the flexibility of the origami plates, structures that are otherwise inherently non-foldable can achieve large deformations and overall motions, such as the magic spiral cube (MSC) origami structures. Modeling such structures requires developing a novel high-fidelity numerical model capable of simulating complex plate deformations, large rotational motions, and predicting nonlinear force-displacement relationships. Thus, the MSC origami structure is modeled as flexible plates rotating about creases. The positions and position-gradient vectors at the discretized nodes serve as generalized coordinates to describe arbitrary displacements. The crease effect is equivalently modeled as torsional springs characterized by plate normal, expressed as generalized forces acting on plate vertices through the principle of virtual work. A dynamic numerical model for flexible multibody systems is established using the absolute nodal coordinate formulation. Considering the challenge of determining physical parameters in the numerical model, a neural network-driven weighted hybrid framework is adopted to identify the required key parameters. Following parameter calibration through experimental measurements, the dynamic response of a simple double-plate origami and the folding motion of the MSC origami structure are simulated and experimented. Quantitative comparison results demonstrate the model’s consistency, validity, and accuracy in predicting nonlinear dynamics of flexible origami structures, proving its capability to capture large overall rotations and deformations. This work holds significant application potential in accurate numerical computation, mechanical performance prediction, and structural optimization design for flexible MSC origami structures.</p>