Simulated outgoing longwave radiation (OLR) outputs by two versions of the grid-point atmospheric general circulation model (GAMIL) were analyzed to assess the influences of improvements in cloud microphysics and convective parameterization schemes on the simulation of the Madden-Julian oscillation (MJO) and other tropical waves. The wavenumber-frequency spectral analysis was applied to isolate dominant modes of convectively coupled equatorial waves, including the MJO, Kelvin, equatorial Rossby (ER), mixed Rossby-gravity (MRG), and inertio-gravity (IG) waves. The performances of different versions of the GAMIL model (version 1.0 (GAMIL1.0) and version 2.0 (GAMIL2.0)) were evaluated by comparing the power spectrum distributions of these waves among GAMIL1.0, GAMIL2.0, and observational data. GAMIL1.0 shows a weak MJO signal, with the maximum variability occurring separately at wavenumbers 1 and 4 rather than being concentrated on wavenumbers 1–3, suggesting that GAMIL1.0 could not effectively capture the intraseasonal variability. However, GAMIL2.0 is able to effectively reproduce both the symmetric and anti-symmetric waves, and the significant spectra of the MJO, Kelvin, and MRG waves are in agreement with observational data, indicating that the ability of GAMIL2.0 to simulate the MJO and other tropical waves is enhanced by improving the cloud microphysics and convective parameterization schemes and implying that such improvements are crucial to further improving this model’s performance.