Abstract: To address the problems encountered in the forward and reverse prediction of wellbore pressure distribution for multiphase flow in oil and gas engineering, such as strong parameter coupling, low iteration efficiency of traditional algorithms, and degraded accuracy under complex working conditions, this paper proposes a quantum-classical neural network prediction model for wellbore pressure distribution based on a discrete-variable (DV) quantum circuit. Forward and reverse prediction models are constructed for bottomhole pressure and wellhead pressure, respectively. First, the Beggs‑Brill correlation is adopted to generate 10, 000 valid wellbore multiphase flow pressure datasets with 15‑dimensional parameters covering wellhead working conditions, pipe geometric characteristics, and fluid physical properties. Then, a discrete‑variable quantum‑classical neural network is designed, consisting of a classical preprocessor, a discrete‑variable quantum core, and a classical postprocessor. The classical network is used to reduce feature dimensionality, and the superposition and entanglement characteristics of the DV quantum circuit are exploited to capture the deep nonlinear coupling features among parameters. The classical postprocessor then maps the output of the quantum circuit to pressure values. Joint optimization and training of the parameters in both the classical and quantum networks are realized via error backpropagation using the 10, 000 samples. Furthermore, the trained quantum‑classical neural network surrogate model is applied to wellbore‑reservoir coupled simulation, which requires frequent calculation of wellbore multiphase flow pressure distribution. Comparative tests with traditional coupled models systematically verify the practical performance of the proposed surrogate model in terms of computational efficiency and prediction accuracy. Numerical results show that both the bottomhole pressure forward prediction model and the wellhead pressure reverse prediction model achieve stable convergence of training errors. In validation on new samples, their relative errors are as low as 0.16% and 0.18%, respectively. Even under extreme edge conditions such as large negative inclination and ultra‑long pipelines, no degradation in prediction accuracy is observed. In practical wellbore‑reservoir coupled simulation, compared with the traditional coupled simulation method, the relative errors of bottomhole flowing pressure are only 4.177% for production wells and 2.863% for injection wells, while the relative errors of flow rate are 1.7% and 1.69%, respectively. Meanwhile, the computational efficiency is improved by nearly five times. In summary, the discrete‑variable quantum‑classical neural network proposed in this paper effectively achieves high‑precision and high‑stability prediction of wellbore pressure. It provides a high‑performance, high‑reliability technical pathway for wellbore pressure calculation and wellbore‑reservoir coupled simulation in the oil and gas engineering field in the quantum computing era.