Response surface methodology (RSM), Artificial Neural Network (ANN) and Radial Basis Function Neural Network (RBFNN) were applied to model and predict the efficiency of two carcinogenic dyes (Methylene blue (MB) and Malachite green (MG)) adsorption onto Mn@ CuS/ZnS nanocomposite-loaded activated carbon (Mn@ CuS/ZnS-NC-AC) as a novel adsorbent. The properties of Mn@ CuS/ZnS-NC-AC were identified by XRD; FE-SEM and EDS. The parameters such as pH, Mn@ CuS/ZnS-NC-AC mass, sonication time, MB concentration and MG concentration involved in the adsorption process were set within the ranges 4.0–8.0, 0.010–0.030 g, 1–5 min, 5–25 mg L⁻¹ and 5–25 mg L⁻¹, respectively. The applicability of the RBFNN, ANN and RSM models for the description of experimental data was examined using four statistical criteria (coefficient of determination (R²), root mean square error (RMSE), mean absolute error (MAE) and absolute average deviation (AAD)). Compared to RSM, the RBFNN and ANN exhibited better performance for modeling the process of both dyes adsorption. The significant factors were evaluated followed by the optimization of the process. The adsorption of MB and MG was found to be mostly affected by the concentration of MB and MG dyes. The equilibrium adsorption data were analyzed by Langmuir, Freundlich, Temkin and Dubinin–Radushkevich isotherm models. The best fit to the data was obtained by applying the Langmuir model. Meanwhile, the maximum adsorption capacity for MB and MG was estimated to be 126.42 and 115.08 mg g⁻¹, respectively.