Climate change is altering hydrology, land cover, and biogeochemistry in Alpine river systems, yet predictive understanding of dissolved organic carbon (DOC) and total suspended solids (TSS), across glacierised and lowland catchments remains limited. This knowledge gap constrains our ability to forecast impacts on carbon cycling and sediment management. We present a data-driven predictive model for Swiss streams from diverse catchments, spanning glacierised high-mountain basins to lowland agricultural and forested systems. The machine learning framework incorporates discharge, water quality, and land use and land cover changes to predict DOC and TSS, with high accuracy (RMSE=14% of standard deviation for DOC) following validation of the best performing algorithm. While its reliance on routinely measured parameters makes it adaptable for near real-time forecasting, the model was designed for climate change scenario analysis. Projections indicate that by 2090, under RCP8.5, DOC exports will rise by ∼50% in high-mountain catchments and ∼15% in lowland systems, primarily driven by discharge, not by land cover change. TSS responses vary seasonally and by catchment, with increases in many glacierised basins and decreases in most lowland streams. Seasonal DOC load peaks are projected to occur earlier in the year. By harmonising diverse datasets and quantifying site-specific climate, hydrology and land cover interactions, this approach provides a tool for managing carbon and sediment fluxes in rapidly changing Alpine environments.