We examine the effectiveness of recursive-weight and equal-weight combination strategies for forecasting using many time-varying models of the relationship between inflation and the output gap. The forecast densities for inflation reflect the uncertainty across models using many statistical measures of the output gap, and allow for time-variation in the ensemble Phillips curves. Using real-time data for the US, Australia, New Zealand and Norway, we find that the recursive-weight strategy performs well, consistently giving well-calibrated forecast densities. The equal-weight strategy generates poorly-calibrated forecast densities for the US and Australian samples. There is little difference between the two strategies for our New Zealand and Norwegian data. We also find that the ensemble modelling approach performs more consistently with real-time data than with revised data in all four countries.