NASA and IBM have jointly launched Surya, an open-source artificial intelligence model designed to forecast solar weather with enhanced precision. Surya, available on the Hugging Face platform, is trained on nine years of high-resolution imagery data from NASA's Solar Dynamics Observatory, totaling approximately 218 terabytes. The model leverages a 366 million parameter transformer architecture to predict solar flares and geomagnetic storms up to two hours in advance, improving solar flare classification accuracy by 16% compared to previous methods. Surya aims to provide earlier warnings for potentially disruptive solar storms, including Carrington-class events, thereby safeguarding critical technology and infrastructure. The AI model also supports better forecasting of auroras and contributes to heliophysics research. The development was led by the IMPACT AI team at NASA Marshall Space Flight Center, emphasizing collaboration between NASA's extensive solar data and IBM's AI expertise. By open-sourcing Surya, the organizations intend to enable researchers worldwide to advance space weather prediction and protection strategies.