System simulation of high-pressure liquid–gas ejectors requires compact pressure–flow representations. This work compares two reduced-order models of the same fixed-geometry water–methane ejector against a 21-point CFD pressure-response database: an effective-area onedimensional model and an explicit quadratic surrogate derived from the CFD data. With leave-one-out errors normalized by the global CFD response range, the surrogate gives NRMSEs of 1.19%, 0.80%, and 0.71% for primary flow, secondary flow, and entrainment ratio; the 1D model gives 1.31%, 5.49%, and 5.52%. Both models are implemented in Modelica and exported through FMI 2.0. The surrogate CS FMU matches the native Modelica scheduling trajectory to machine precision. Coupled to a first-order back-pressure actuator, the surrogate supports inverse entrainment-ratio scheduling with settling times of 7.0 and 7.6 s. The two formulations therefore offer complementary choices for reusable system simulation.