Accurate precipitation (P) estimates are critical for effective rainfall-runoff modeling in hydrological studies. However, achieving this accuracy can be particularly challenging in complex terrains, where much of the area is ungauged or poorly gauged. This study investigates the performance and hydrological utility of the GPM (IMERG_F) satellite precipitation dataset (SPD) for predicting daily streamflow in the mountainous Upper Jhelum River Basin (UJRB), Pakistan. The analysis employs the SWAT hydrological model alongside SWAT integrated with soft computing models (SCMs), including Random Forests (SWAT-RF), Extreme Gradient Boosting (SWAT-XGBoost), and Long Short-Term Memory networks (SWAT-LSTM). SCMs were trained using outputs from uncalibrated SWAT models to enhance prediction accuracy. Among these models, SWAT-XGBoost demonstrated the highest performance during both training and testing phases, with R² values of 0.98 and 0.96, respectively. This outperformed SWAT-RF (R² = 0.97 and 0.94), SWAT-LSTM (R² = 0.90 and 0.85), and SWAT-CUP (R² = 0.70 and 0.72). These results suggest that hydrological SCMs coupled with GPM SPD represent a promising approach for streamflow simulation, especially in regions with sparse or uneven gauge networks.