NASA and IBM have jointly developed and launched Surya, an open-source foundation AI model designed to improve the forecasting of solar flares and space weather events. Surya utilizes a transformer architecture with 366 million parameters and has been trained on an extensive dataset comprising nine years and 218 terabytes of high-resolution atmospheric satellite images. This model enhances the accuracy of solar flare classification by 16% compared to previous methods and can predict solar activity up to two hours in advance. The improved forecasting capabilities offered by Surya aim to provide earlier warnings for geomagnetic storms, which can impact satellites, electrical grids, and telecommunications infrastructure. The AI model is expected to aid NASA, the National Oceanic and Atmospheric Administration (NOAA), and other scientific entities in better anticipating solar storms and auroras, potentially helping to mitigate the effects of severe space weather events such as Carrington-class solar storms. Surya was developed by the IMPACT AI team at NASA's Marshall Space Flight Center and is available on the Hugging Face platform for broader scientific use.