Ants are inherently social insects whose observed feeding behavior (i.e., trophallaxis events) can be represented as a dynamic, high frequency contact network. Traditional statistical analyses of network models often aggregate data over time. However, in doing so, the information contained in data observed at high frequencies may be lost. We present a continuous-time dynamic network model based on a continuous-time Markov process, with states being defined by the network topology, and transition rates modeled as a function of individual (node-specific) and pairwise (edge-specific) covariates measured over time. We demonstrate this model on data consisting of observed trophallaxis events among a colony of common black carpenter ants, collected over 14400 seconds.