Inter-turn short circuits are a major cause of winding deformation, insulation degradation, and failure in power transformers. An efficient and accurate simulation model is essential to analyze the evolution of electrical, magnetic, structural, and acoustic characteristics under such faults, which is crucial for revealing performance deterioration mechanisms. However, the strong nonlinear high-frequency transient shocks from short-circuit currents degrade the convergence and stability of traditional field-circuit coupled models, making it hard to balance accuracy and efficiency. To address this issue, this paper proposes a reduced-order hybrid modeling and efficient computation method for transformer multi-physics under inter-turn short circuits. First, based on the multi-conductor transmission line model, analytical expressions for winding voltage, winding current, and circulating current are derived to enable rapid excitation calculation. Second, a reduced-order hybrid strategy combining analytical and finite element models is established, enabling efficient electromagnetic-structural-acoustic co-simulation while reducing complexity. Finally, experiments on a 110 kV oil-immersed transformer with inter-turn short circuits are conducted to validate the proposed method. Results show that the current error is only 2.51%. Compared with the traditional fully coupled model, the proposed method improves the sound pressure spectrum evaluation metrics, with PCC increased by 18.6% and RMSE and JSD reduced by 22.6% and 27.8%, respectively. Meanwhile, memory usage is reduced by 37.1% and computational efficiency is improved by 29.8%. These findings confirm the proposed method's advantages in accuracy and efficiency, providing a reliable modeling basis for online monitoring and intelligent diagnosis.