 This paper presents a novel approach to accurately predict gross primary productivity, GPP, on the Tibetan Plateau using a deep learning model. The model was trained using data from nine flux observation sites and incorporated environmental factors such as temperature, photosynthetic active radiation, enhanced vegetation index, normalized difference vegetation index, and leaf area index. The model was found to be highly accurate, achieving an overall R2 value of 0.870, root mean square error, RMSC, of 0.788 GcM-2d-1, and mean absolute error, MAE, of 0.440 GcM-2d-1. Additionally, the model showed that distance and vegetation type were both influential factors in GPP prediction. The site located in the alpine swamp meadow was insensitive to changes in environmental factors, while the site located in the alpine cobrasia meadow had greater sensitivity. Finally, this study demonstrates that deep learning can provide a reliable tool for GPP prediction, taking into account spatial, temporal, and environmental factors. This article was authored by Qin Mengyang, Ning Mengya, Yan Gangwang, and others. We are article.tv, links in the description below.