Context. Luminous Galactic OH/IR stars have been identified as massive (>4-5 M⊙) asymptotic giant branch (AGB) stars experiencing hot bottom burning and Li production. Their Rb abundances and [Rb/Zr] ratios, as derived from classical hydrostatic model atmospheres, are significantly higher than predictions from AGB nucleosynthesis models, posing a problem for our understanding of AGB evolution and nucleosynthesis. Aims: We report new Rb and Zr abundances in the full sample (21) of massive Galactic AGB stars, previously studied with hydrostatic models, by using more realistic extended model atmospheres. Methods: For this, we use a modified version of the spectral synthesis code Turbospectrum and consider the presence of a circumstellar envelope and radial wind in the modelling of the optical spectra of these massive AGB stars. The Rb and Zr abundances are determined from the 7800 Å Rb I resonant line and the 6474 Å ZrO bandhead, respectively, and we explore the sensitivity of the derived abundances to variations of the stellar (Teff) and wind (Ṁ, β and vexp) parameters in the pseudo-dynamical models. The Rb and Zr abundances derived from the best spectral fits are compared with the most recent AGB nucleosynthesis theoretical predictions. Results: The Rb abundances derived with the pseudo-dynamical models are much lower (in the most extreme stars even by 1-2 dex) than those derived with the hydrostatic models, while the Zr abundances are similar. The Rb I line profile and Rb abundance are very sensitive to the wind mass-loss rate Ṁ (especially for Ṁ ≥ 10-8M⊙ yr-1) but much less sensitive to variations of the wind velocity-law (β parameter) and the expansion velocity vexp(OH). Conclusions: We confirm the earlier preliminary results based on a smaller sample of massive O-rich AGB stars, suggesting that the use of extended atmosphere models can solve the discrepancy between the AGB nucleosynthesis theoretical models and the observations of Galactic massive AGB stars. The Rb abundances, however, are still strongly dependent on the wind mass-loss Ṁ, which, unfortunately, is unknown in these AGB stars. Accurate mass-loss rates Ṁ (e.g. from rotationally excited lines of the CO isotopologues in the radio domain) in these massive Galactic AGB stars are needed in order to break the model's degeneracy and obtain reliable (non-model-dependent) Rb abundances in these stars.