NASA and IBM Launch Open-Source Lunar Foundation Model

NASA and IBM have launched the NASA-IBM Lunar Foundation Model, an open-source AI model available for free. The announcement was made on 2026-09-10.

The announcement date was 2026-09-10. NASA and IBM developed the AI model together. NASA's partnership includes ongoing collaboration with IBM Research and several academic institutions, and NASA is developing and releasing science-focused AI models with IBM.

The NASA-IBM Lunar Foundation Model was launched as an open-source AI model and is available for free. It is hosted publicly on Hugging Face, and its complete codebase is available on GitHub.

The model was trained primarily on data from NASA's Lunar Reconnaissance Orbiter (LRO), and the AI model is mostly trained on LRO data. LRO collected 17 years of data and produced an almost complete mosaic of images.

The model's specifications include roughly 2,000,000 image tiles, more than 1,000,000 high-resolution camera images, 1-meter resolution, 964,000 multispectral images, and 100-meter resolution.

Additional training used data from GRAIL, Lunar Prospector, and Selenological and Engineering Explorer.

Its tasks include mapping craters, identifying young volcanic features, and modeling possible water-ice locations near the lunar poles.

In the Moon's permanently shadowed regions, ice can remain preserved for as long as billions of years.

The release makes a lunar AI model freely available, with public hosting and full code. Its training draws on LRO and other lunar mission data, while its listed tasks cover crater mapping, young volcanic feature detection, and modeling possible water-ice locations near the lunar poles.