We propose a multimodel ensemble forecast framework for streamflow forecasts at multiple locations that incorporates large-scale climate information. It has four broad steps: (1) Principal component analysis is performed on the spatial streamflows to identify the dominant modes of variability. (2) Potential predictors of the dominant streamflow modes are identified from several large-scale climate features and snow water equivalent information. (3) Objective criterion is used to select a suite of candidate nonlinear regression models each with different predictors. (4) Ensemble forecasts of the dominant streamflow modes are generated from the candidate models and are combined objectively to produce a multimodel ensemble, which are then back transformed to produce spatially coherent streamflow forecasts at all the locations. The utility of the framework is demonstrated in the skillful forecast of spring seasonal streamflows at six locations in the Gunnison River Basin at several lead times. The generated ensemble streamflow forecast provides valuable and useful information for optimal management and planning of water resources in the basin.