Context. Mass loss plays a crucial role in the lives of massive stars, especially as the star leaves the main sequence and evolves to the red supergiant (RSG) phase. Despite its importance, the physical processes that trigger mass-loss events in RSGs are still not well understood. Recently, we showed that adding a semi-empirical wind to atmosphere models can accurately reproduce observed extensions in the atmospheres of RSGs, where the mass-loss events are taking place, particularly in the CO and water lines. Aims. By adding a static wind to a MARCS atmospheric model, we computed synthetic observables that match new interferometric data of the RSGs AH Sco, KW Sgr, V602 Car, CK Car, and V460 Car obtained with the VLTI/MATISSE and VLTI/GRAVITY instruments between August 2022 and February 2023. We also used archival VLTI/AMBER data of KW Sgr and VLTI/GRAVITY data of AH Sco. The MATISSE wavelength range includes the SiO molecule at 4.0 μm with a spectral resolution of R ~ 500. Methods. The model intensities with respect to the line-of-sight angle (μ) as well as the spectra and visibilities were computed using the stellar radiative transfer code TURBOSPECTRUM. We found the best-fit model, mass-loss rate, and best-fit angular Rosseland diameter for the observations. We simultaneously matched our model to the data, covering a wavelength range of 1.8–5.0 μm, which corresponds to the K, L, and M bands. Results. Our models reproduce the spectro-interferometric data over this wide wavelength range, including extended atmospheric layers of CO, H2O, and SiO. We obtain a range of Rosseland angular diameters between 3.0 < θRoss < 5.5 mas and a range of mass-loss rates of −6.5 < log Ṁ/M⊙ yr−1 < −4 for our five targets. In our best-fit models, the partial pressure of SiO relative to the gas pressure, PSiO/Pg, and the SiO 4.0 μm line intensity increase between 2 and 3 stellar radii. The relative intensity depends on the luminosity used for our models, since the more luminous models have a higher mass-loss rate. Conclusions. This work further demonstrates that our MARCS+wind model can reproduce the observed physical extension of RSG atmospheres for several spectral diagnostics spanning a broad wavelength range. We reproduce both spectra and visibilities of newly obtained data as well as provide temperature and density stratifications that are consistent with the observations. With the MATISSE data, we newly include the extension of SiO layers as a precursor of silicate dust.