Glacier retreat caused by global warming alters the hydrological regime and poses far-reaching challenges to water resources and nature conservation of the headwater of Yangtze River, and its vast downstream regions with dense population. However, there is still lack of a robust modeling framework of the “climate-glacier-streamflow” in this water tower region, to project the future changes of glacier mass balance, glacier geometry, and the consequent impacts on runoff. Moreover, it is imperative to use the state-of-the-art sixth phase Coupled Model Intercomparison Project (CMIP6) to assess glacio-hydrology variations in future. In this study, we coupled a glacio-hydrological model (FLEX^G) with a glacier retreat method (Δh-parameterization) to simulate glacio-hydrological processes in the Dongkemadi Glacier (over 5155 m.a.s.l), which has the longest continuous glacio-hydrology observation on the headwater of Yangtze River. The FLEX^G-Δh model was forced with in-situ observed meteorological data, radar ice thickness, remote sensing topography and land cover data, and validated by measured runoff. The results showed that the model was capable to simulate hydrological processes in this glacierized basin, with Kling-Gupta efficiency (I_KGE) of daily runoff simulation 0.88 in calibration and 0.70 in validation. Then, forcing by the bias-corrected meteorological forcing from the eight latest CMIP6 Earth system models under two climate scenarios (RCP2.6 and RCP8.5), we assessed the impact of future climate change on glacier response and its hydrological effects. The results showed that, to the end of simulation in 2100, the volume of the Dongkemadi Glacier would continuously retreat. For the RCP2.6 and RCP8.5 scenarios, the glacier volume will decrease by 8.7 × 10⁸ m³ (74%) and 10.8 × 10⁸ m³ (92%) respectively in 2100. The glacier runoff will increase and reach to peak water around 2060 to 2085, after this tipping point water resources will likely decrease.