# IBM and NASA opened Surya for heliophysics research, with a NASA technical record of its solar inputs IBM announced Surya on August 20, 2025 as an open heliophysics foundation model developed with NASA for work with solar-observation data. Canonical: https://rangoon.ai/insights/archive-ibm-nasa-surya-heliophysics-august-2025/ Author: Rangoon Editorial (https://rangoon.ai/insights/editorial/) Published: 2026-10-24 Updated: 2026-10-24 Event date: 2025-08-20 Tags: [IBM](https://rangoon.ai/insights/tags/ibm/), [NASA](https://rangoon.ai/insights/tags/nasa/), [space science](https://rangoon.ai/insights/tags/space-science/), [open models](https://rangoon.ai/insights/tags/open-models/) Source note: Historical reporting: IBM announced Surya on August 20, 2025 as a heliophysics foundation model developed with NASA and made available through public model and code channels. NASA Technical Reports Server lists the related preprint’s publication date as August 20, while separately recording August 18 as its acquisition date; the record documents Surya’s 366 million parameters and SDO input channels. The sources establish an open research release and technical model description; they do not establish an operational space-weather warning system, real-time forecasting deployment, or improved warning performance. Image: Solar-data research model. Original Rangoon explanatory diagram: Source boundary: dated primary release reporting; original editorial analysis. No vendor interface, photograph, measurement result, or endorsement implied. ## Key takeaways - IBM announced an open Surya heliophysics model developed with NASA. - NASA’s report documents the model size and solar-observation input channels. - Open availability for research is distinct from an operational space-weather forecast service. ## Surya was introduced as an open heliophysics research model [source 1](https://research.ibm.com/blog/surya-heliophysics-ai-model-sun) IBM introduced Surya on August 20, 2025 as a new heliophysics foundation model developed with NASA. The company said the model was available through public code and model channels, making this a release event rather than a report of a new operational warning service. Heliophysics concerns the Sun and its interaction with the space environment, a setting where scientific models can support research without automatically becoming decision systems. IBM’s announcement describes possible uses and access routes, but it does not say that Surya had taken responsibility for real-time solar alerts or that an agency had adopted it as an operational forecast baseline. The distinction is material: a public model can enable experimentation and later validation while remaining separate from production pipelines that have defined monitoring, reliability and human-review requirements. ## NASA’s report documents inputs and scale but does not substitute for an operational evaluation [source 2](https://ntrs.nasa.gov/citations/20250008498) NASA’s technical record for Surya describes a 366 million-parameter model and identifies multichannel Solar Dynamics Observatory inputs, including eight Atmospheric Imaging Assembly channels and five Helioseismic and Magnetic Imager channels. This gives the archive a concrete technical description from a NASA record rather than relying only on a company blog. It does not disclose every training and deployment detail that would be needed to predict performance in a specific operational task. Input channels capture different measurements, but their presence does not demonstrate that all future solar phenomena are forecast accurately. The report is also not a public service-level agreement, a warning-accuracy study or a comparison across all competing heliophysics systems. It supports a bounded description of the released research model’s documented scale and inputs. ## Open code and weights widen inspection and reuse, while users still own task validation [source 1](https://research.ibm.com/blog/surya-heliophysics-ai-model-sun) [source 2](https://ntrs.nasa.gov/citations/20250008498) IBM described Surya as open source and pointed to public release channels. Open availability can make it easier for other researchers to inspect implementation choices, test a model on new data and develop specialised variants. It cannot transfer responsibility for validation from a prospective user to the original release. A team adapting solar imagery for a particular event-detection or forecasting task would still need to define labels, time boundaries, error costs, baselines and a way to handle uncertainty. The sources do not say that all intended applications have passed those steps. They also do not show that a model output directly predicts a geomagnetic consequence at Earth, because that depends on physical processes and data beyond a single imagery representation. The event supports an open-research story, not a completed operational deployment. ## A foundation model release is not evidence of a new space-weather safety or warning capability [source 1](https://research.ibm.com/blog/surya-heliophysics-ai-model-sun) [source 2](https://ntrs.nasa.gov/citations/20250008498) The source materials make a strong, limited claim: IBM and NASA made a heliophysics foundation model publicly available and documented its solar-data inputs. They do not establish that Surya improved the timeliness or accuracy of a public space-weather warning, reduced risk to spacecraft, or met a named operational standard. No source provides a real-time incident record, a false-alarm analysis, a human-in-the-loop procedure or a reliability target. Scientific usefulness also varies by task and data regime. The archive therefore preserves the release and technical description while refusing a common leap from foundation-model availability to safety-critical deployment. A later operational agency record, tied to a named system and measured evaluation period, would be necessary to show that transition. ## Sources 1. [IBM Research: Introducing Surya, a new heliophysics foundation model](https://research.ibm.com/blog/surya-heliophysics-ai-model-sun) 2. [NASA Technical Reports Server: Surya: Foundation Model for Heliophysics](https://ntrs.nasa.gov/citations/20250008498) ## Related reading - [NASA and IBM launched a lunar foundation model with code and model access for science work](https://rangoon.ai/insights/archive-nasa-ibm-lunar-foundation-model-september-2026/) - [IBM released Granite-Docling-258M as a compact document-conversion VLM](https://rangoon.ai/insights/archive-ibm-granite-docling-258m-september-2025/) - [Ultralytics released YOLO26 with optional NMS-free real-time vision inference](https://rangoon.ai/insights/archive-ultralytics-yolo26-vision-release-january-2026/) ## More information - [Documentation index](https://rangoon.ai/llms.txt) - [AI access and policies](https://rangoon.ai/ai/) - [Source repository](https://github.com/hypler-dev/rangoon)