The ability to predict the mechanical properties of organic semiconductors is of critical importance for roll-to-roll production and thermomechanical reliability of organic electronic devices. Here, we describe the use of coarse-grained molecular dynamics simulations to predict the density, tensile modulus, Poisson ratio, and glass transition temperature for poly(3-hexylthiophene) (P3HT) and its blend with C-60. In particular, we show that the resolution of the coarse-grained model has a strong effect on the predicted properties. We find that a one-site model, in which each 3-hexylthiophene unit is represented by one coarse-grained bead, predicts significantly inaccurate values of density and tensile modulus. In contrast, a three-site model, with one coarse-grained bead for the thiophene ring and two for the hexyl chain, predicts values that are very close to experimental measurements (density = 0.955 g cm-3, tensile modulus = 1.23 GPa, Poisson ratio = 0.35, and glass transition temperature = 290 K). The model also correctly predicts the strain-induced alignment of chains as well as the vitrification of P3HT by C-60 and the corresponding increase in the tensile modulus (tensile modulus = 1.92 GPa, glass transition temperature = 310 K). We also observe a decrease in the radius of gyration and the density of entanglements of the P3HT chains with the addition C-60 which may contribute to the experimentally noted brittleness of the composite material. Although extension of the model to poly(3-alkylthiophenes) (P3ATs) containing side chains longer than hexyl groups nonyl (N) and dodecyl (DD) groups correctly predicts the trend of decreasing modulus with increasing length of the side chain measured experimentally, obtaining absolute agreement for P3NT and P3DDT could not be accomplished by a straightforward extension of the three-site coarse-grained model, indicating limited transferability of such models. Nevertheless, the accurate values obtained for P3HT and P3HT:C-60 blends suggest that coarse graining is a valuable approach for predicting the thermomechanical properties of organic semiconductors of similar or more complex architectures.