Statistical timing characterization for modeling On-Chip Variation (OCV) of library cells is critical in current technology nodes to avoid over-design and to improve design convergence and predictability. OCV characterization is, however, resource intensive as it involves running millions to billions of Monte-Carlo spice simulations to build the entire standard cell timing library. In this paper, we leverage a deep learning technique to model cell propagation delays as well as OCV sigma at target process-voltage-temperature (PVT) corners with a significantly reduced number of simulations of 60% while also achieving a high accuracy of 99.98%.