A hybrid stacking model is proposed for enhanced short-term load forecasting (STLF) by integrating multiple AI models with Lasso regression. The base models include ANN, XgBoost, LSTM, Stacked LSTM, and Bi-LSTM, while Lasso regression serves as the metalearner. The model considers factors such as temperature, rainfall, and electricity prices to improve prediction accuracy. Empirical analyses on real-world datasets from Australia and Spain show significant improvements in forecasting accuracy, with a substantial reduction in MAPE compared to existing models. The stacking technique effectively combines the strengths of different models, enhancing predictive performance. The model outperforms five single AI models and two hybrid models, demonstrating superior accuracy in both stable and unstable load conditions. The study highlights the efficiency of the stacking technique in improving STLF accuracy, offering potential benefits for the power industry. The model integrates external factors to simulate real-world conditions, enhancing its applicability. The proposed model shows strong performance in identifying seasonal patterns and load trends, with significant improvements in MAPE compared to other models. The study also discusses the model's economic and power grid implications, emphasizing its potential for cost savings and grid stability. Limitations include challenges in predicting minimum values and partial load prediction in unstable environments. Future work could involve incorporating more variables and testing with larger datasets. The model's effectiveness is validated through case studies, demonstrating its potential for proactive power planning and improved grid reliability.A hybrid stacking model is proposed for enhanced short-term load forecasting (STLF) by integrating multiple AI models with Lasso regression. The base models include ANN, XgBoost, LSTM, Stacked LSTM, and Bi-LSTM, while Lasso regression serves as the metalearner. The model considers factors such as temperature, rainfall, and electricity prices to improve prediction accuracy. Empirical analyses on real-world datasets from Australia and Spain show significant improvements in forecasting accuracy, with a substantial reduction in MAPE compared to existing models. The stacking technique effectively combines the strengths of different models, enhancing predictive performance. The model outperforms five single AI models and two hybrid models, demonstrating superior accuracy in both stable and unstable load conditions. The study highlights the efficiency of the stacking technique in improving STLF accuracy, offering potential benefits for the power industry. The model integrates external factors to simulate real-world conditions, enhancing its applicability. The proposed model shows strong performance in identifying seasonal patterns and load trends, with significant improvements in MAPE compared to other models. The study also discusses the model's economic and power grid implications, emphasizing its potential for cost savings and grid stability. Limitations include challenges in predicting minimum values and partial load prediction in unstable environments. Future work could involve incorporating more variables and testing with larger datasets. The model's effectiveness is validated through case studies, demonstrating its potential for proactive power planning and improved grid reliability.