Dryland vegetation is highly sensitive to climate change and human activities and is exhibiting frequent degradation and recovery. Accurately quantifying its spatiotemporal changes and driving mechanisms is crucial for ensuring ecological conservation and sustainable management. In this study, interpretable machine learning is employed to analyze the nonlinear response of the dryland growing season vegetation gross primary productivity (GPP) to climate factors and to identify key thresholds across China and Mongolia from 1982 to 2022. Additionally, residual trend analysis is used to quantify the relative contributions of climate change and human activities to GPP variations. The results indicate a general increasing trend in GPP across the study area, averaging 0.26 gCm- 2month- 1, with its spatial distribution shaped by the aridity gradient. The GPP increases more slowly in more arid regions. While the dryland vegetation GPP in Mongolia is greater than that in China, it has exhibited more severe degradation, whereas dryland vegetation in China has gradually improved. Temperature and precipitation emerge as the dominant climate factors influencing the GPP, with their critical thresholds varying along the aridity gradient. Climate factors primarily drive GPP changes, whereas human activities exert a greater influence on vegetation dynamics in marginal zones. The combined effects of climate change and human activities accounted for 85.78 % of vegetation GPP changes across the study area. Precipitation-dominated climate regimes play crucial roles in shaping GPP trends in the drylands of China and Mongolia. These findings provide valuable insights for cross-border ecological restoration, climate adaptation strategies, and global dryland vegetation management.