The cerebral cortex must flexibly alternate between locally segregated activity that supports specialization and long-range interactions that enable integration. How cortical networks balance these competing demands remains unclear. We propose that fluctuations around a critical point between ordered and disordered phases provide a natural framework to understand coordinated neuronal activity across scales. Using simultaneous spiking recordings from the left and right prefrontal cortex (PFC) of freely behaving rats, we found that signatures of criticality emerged at the onset of neuronal synchronization, both locally within individual hemispheres and globally across the combined population. Over time, cortical activity spanned a continuum of intra- and interhemispheric synchrony, including states in which neurons were locally desynchronized yet maintained functional interhemispheric coordination. A computational model operating near its critical regime reproduced these empirical results. When slow fluctuations in local excitability were introduced, the model generated dynamic, activity-dependent interpopulation correlations despite fixed coupling between populations. These results suggest that cortical networks achieve flexible transitions between local and global computation by fluctuating around a critical regime.