The recent development of multi-¬‐model ensemble (MME) databases containing real-¬time and historical forecasts, such as the North American Multi-¬‐model Ensemble (NMME) and the WCRP/WWRP Subseasonal to Seasonal (S2S) project, provides an ideal opportunity to explore predictability and prediction on subseasonal to seasonal timescales. Furthermore, the multi-¬‐model ensemble approach has been shown to have superior average skill to any single model. We invite contributions that explore predictability and prediction using multiple models, including initialization strategies, optimal combination strategies, methods for calibration and correcting biases, skill assessments, predictability studies, understanding sources of skill, and identifying forecasts of opportunity.