Elastin-like polypeptides (ELPs) undergo a sharp solubility transition from low temperature solvated phases to coacervates at elevated temperatures, driven by the increased strength of hydrophobic interactions at higher temperatures. The transition temperature, or 'cloud point', critically depends on sequence composition, sequence length, and concentration of the ELPs. In this work, we present a temperaturedependent, implicit solvent, sequence-specific coarse-grained (CG) simulation model that reproduces the transition temperatures as a function of sequence length and guest residue identity of various experimentally probed ELPs to appreciable accuracy. Our model builds upon the self-organized polymer model introduced recently for intrinsically disordered polypeptides (SOP-IDP), and introduces a semi-empirical functional form for the temperature-dependence of hydrophobic interactions. In addition to the fine performance for various ELPs, we demonstrate the ability of our model to capture the thermal compactions in dominantly hydrophobic intrinsically disordered polypeptides (IDPs), consistent with experimental scattering data. With the high computational efficiency afforded by the CG representation, we envisage that the model will be ideally suited for simulations of large-scale structures such as ELP networks and hydrogels, as well as agglomerates of IDPs.