<p>Seasonal rainfall prediction for the Indian Summer Monsoon remains a challenging task for climate models, with regional-scale forecasts proving even more difficult than predicting the all-India averaged seasonal rainfall. In this study, we evaluate the performance of eight models from the Copernicus Climate Change Service (C3S) in simulating and predicting monsoon rainfall over different homogeneous regions in India. Our analysis reveals complex regional variations in prediction skill, particularly during the June-to-September period. Most C3S models show significantly lower prediction skill for seasonal rainfall over central India, while performing comparatively better for South Peninsular India. Accurate prediction of both monsoon low pressure system (LPS) related rain and non-LPS rain is crucial for forecasting total rainfall in the central Indian region. However, models generally exhibit poor skill in predicting the LPS-related rain as revealed by a weak and statistically insignificant correlation with the observations. The models have a relatively better skill in predicting the non-LPS rainfall over central India. We note that models with good skill in predicting both LPS-related and unrelated rainfall generally exhibit better overall prediction accuracy. Moreover, a realistic representation of teleconnection between the tropical SST and non-LPS rainfall contributes to the skill of the models. Among the eight models analyzed, only the UKMO model demonstrates comparatively good skill in both LPS rain and non-LPS rain, thus providing good predictive ability for rainfall over central India. Conversely, other models either falter in simulating LPS rain or non-LPS rain, consequently compromising their predictive skill for rainfall over central India. Thus, achieving good prediction skills for both LPS rainfall and non-LPS rainfall are deemed essential for accurately forecasting seasonal rainfall over central India.</p>