Influenza poses a significant burden to public health in the US, with millions of illnesses every year. The ability to produce accurate forecasts of public health-relevant targets at actionable geographies have important implications for resource allocation and intervention planning. To gauge forecasting capabilities and guide forecasting efforts, the US Centers for Disease Control and Prevention (CDC) organizes an annual flu forecasting challenge at the national, regional, and state-levels. In this talk I will present Dante, a multiscale flu forecasting model currently participating in the CDC’s 2018 real-time flu forecasting challenge at all geographic scales. Dante is a Bayesian hierarchical model capturing seasonal and geographic trends, inter-seasonal autocorrelations, and heterogenous variability across space and time. State-level forecasts are combined with US Census population data to produce upscaled regional and national forecasts, providing a unified, multiscale framework for flu forecasting. Modeling insights and forecasting results will be presented, as well as comparisons to leading flu forecasting models.