The precision of simulation plays a pivotal role in determining the design parameters of the pressure pipe and distributor in a pneumatic centralized seeding system. This study adopted the discrete element method (DEM) to investigate wheat seed models and their motion characteristics within a pneumatic precision seed-metering device. Using Xinchun No. 6 wheat as the experimental subject, multi-sphere combination models (5, 7, 9, and 11 balls) were employed to describe the seed particle morphology. Moreover, by utilizing the coupling method of the Discrete Element Method (DEM) and Computational Fluid Dynamics (CFD), along with bench tests, the air suspension velocity of seeds and the motion characteristics of the seed-supplying device were analyzed under different particle models. The physical properties of the wheat seeds were measured during the experiments. The simulation results indicated that, as the seed supply rate increased, the airflow velocity distribution within the model became more uniform, enhancing the stability of the suspension velocity. Comparisons between experiments and simulations validated the reliability of the particle models, with the minimum relative error in the suspension velocity determined as 0.21% for the 9-balls model. In addition, compared to the other models, the 9- and 5-balls models more accurately simulated the dynamic behavior of seeds within the seed-supplying device. For the 9-balls model, the relative error of particle velocity in the seed-supplying device is 1.39%, and, in the simulation of displacements in the X and Y directions of the seed-supplying device, the average error is 9.51%. The effectiveness of the multi-sphere combination models was verified, indicating their ability to accurately reflect the dynamic behavior of wheat seeds and improve the design and optimization efficiency of pneumatic precision seed-metering devices.