The Great Lakes of North America form one of the largest freshwater systems on Earth, exhibiting seasonal water level fluctuations that exceed 0.5 meters. These fluctuations pose substantial challenges for coastal resilience, flood risk management, and navigation planning. Accurate seasonal forecasting of lake levels using traditional mechanistic models is challenging due to the complex physical mechanisms and coupled hydroclimatic processes involved. Recently, deep learning has gained prominence in geoscience applications for its ability to recognize intricate patterns within multiphysical datasets. Here, we introduce a novel Dual-Transformer deep learning framework, tested on Lake Superior—the largest of the five Great Lakes. This architecture integrates two modified Transformer models: the Prophet, which predicts underlying trends, and the Critic, which refines the Prophet’s predictions. The final lake level prediction is derived by weighting the outputs of both models through a Multi-Layer Perceptron (MLP), jointly trained with the Prophet and Critic to enhance overall accuracy. Our results demonstrate that the Dual-Transformer model, which uses seven atmospheric and lake features, achieves unprecedented accuracy in seasonal forecasting in the testing dataset, attaining a correlation coefficient of 0.97 and a root mean square error (RMSE) of 4 cm for forecasts up to six months ahead. Additionally, the Dual-Transformer model runs six orders of magnitude faster than conventional mechanistic models, producing results in less than one second on a typical personal computer. These findings suggest our deep learning framework provides an efficient and reliable tool for real-time lake level forecasting, with significant implications for water management and disaster mitigation.