IBM and NASA have unveiled Surya, a cutting-edge open-source AI model for predicting solar weather. Now available on Hugging Face, it empowers global researchers to protect space-based technologies.
 |
| IBM and NASA release Surya, a revolutionary open-source AI model for solar weather forecasting. Accessible on Hugging Face, it supports global research in space technology protection. Image: NASA/ IBM |
Washington DC, USA — August 22, 2025:
In a groundbreaking move to safeguard technological infrastructure, IBM and NASA have launched Surya, an advanced open-source AI model designed to predict solar weather. Surya uses high-resolution solar observation data to provide unprecedented insight into solar activity, crucial for protecting Earth’s space-based technologies, including satellites, power grids, and GPS systems.
Solar flares and coronal mass ejections (CMEs) from the Sun can disrupt vital infrastructure, causing widespread problems such as satellite malfunctions, GPS navigation errors, airline navigational issues, and even power outages. As technology becomes more reliant on space systems, understanding and predicting solar storms has never been more critical.
Juan Bernabe-Moreno, Director of IBM Research Europe, UK, and Ireland, stated: “Think of Surya as a ‘weather forecast for space.’ Just as we prepare for severe weather here on Earth, we must be ready for the impacts of solar storms that could disrupt our modern way of life. This model is a major step forward in securing the future of our technological systems.”
Surya addresses long-standing limitations in solar weather prediction. Traditional methods have been hindered by partial views of the Sun’s surface, making accurate forecasting a challenge. Trained on nine years of solar data from NASA's Solar Dynamics Observatory, Surya breaks these barriers by processing vast datasets to deliver real-time, high-resolution solar predictions.
In early testing, Surya achieved a 16% improvement in solar flare classification accuracy, a notable advancement compared to previous prediction models. For the first time, it can provide visual predictions of solar flares, offering a detailed high-resolution image of flare locations up to two hours before they occur, allowing for proactive mitigation strategies.
The model’s creation involved overcoming significant technical hurdles. Surya was trained on some of the largest and most complex solar observation datasets available, including images that are 10 times larger than typical AI training data. This required a custom multi-architecture solution capable of handling vast amounts of information without compromising efficiency, resulting in a model that offers unparalleled spatial resolution and predictive power.
Kevin Murphy, Chief Science Data Officer at NASA, emphasized: “By integrating NASA’s heliophysics expertise with IBM’s cutting-edge AI, Surya empowers us to analyze solar behavior with exceptional precision, improving our understanding of how solar activity affects critical systems and technologies.”
Surya is part of IBM and NASA’s broader collaboration aimed at advancing scientific research through AI technology. The release on Hugging Face democratizes access to this powerful tool, enabling researchers worldwide to further develop applications tailored to specific industries and regions. This move builds upon previous initiatives, including the release of the Prithvi weather model, which helps predict weather and climate patterns.
As humanity's technological reliance grows, protecting against solar weather disruptions is increasingly vital. Surya is poised to be a game-changer, providing the tools necessary to safeguard infrastructure and ensure the continued success of space exploration.