In the awake human brain, alpha-band activity (8–13 Hz) is the most prominent electrophysiological signature. Several models exist for alpha-rhythm generation using neural mass models, including the Jansen-Rit, Moran-David-Friston, Robinson-Rennie-Wright, and Liley-Wright models [1]. Fitting these models traditionally relies on manual parameter tuning or computationally expensive parameter searches, and they rarely account for inter-individual variation. However, there is substantial experimental variability in the propensity for robust oscillatory activity in EEG, e.g., influenced by structural and genetic factors that are not fully understood. In particular, certain individuals, here termed 'super-oscillators', exhibit a tenfold increase in oscillatory signal-to-noise ratio compared to the population mean, resulting in a high spectral peak in the alpha-band that stands out above the background, 1/f-like activity. Furthermore, in such high-SNR recordings, features like distinct waveform shape with multiple harmonics, as well as variation in amplitude envelope dynamics can be fully observed, whereas they may be obscured in recordings with medium to low signal-to-noise ratio. In this study, we leverage EEG recordings from super-oscillators to estimate parameters of the aforementioned models using simulation-based inference (SBI) [2,3]. SBI utilizes amortized Bayesian inference to efficiently estimate parameters given experimental data, yielding not only point estimates, but parameter distributions that are consistent with the data. We show that this procedure is able to achieve a robust fit of experimental data from different individuals, and in different states. We additionally examine parameter correspondences between different models, as well as highlight where conventional default parameter settings deviate from the empirically derived parameter distributions. Our study lays the groundwork for efficiently improving model selection to advance our understanding of rhythmogenesis of alpha activity. Additionally, it aids in understanding how features derived from macroscale measurements, such as spectral estimates, relate to microscale model parameters, and elucidates factors contributing to inter-individual variability in electrophysiological signals across participants.