# NASA-IBM lunar exploration AI model release

> Live situation record from CLSTR: https://clstr.news/situations/nasa-ibm-lunar-exploration-ai-model-release
> Updated: 2026-10-04T10:10:12.000Z. Sources: 12. Developments: 2.

NASA and IBM have released an open-source artificial intelligence model, known as the NASA-IBM Lunar Foundation Model, to assist in the analysis of lunar surface data. The model is designed to be adaptable to various scientific tasks with minimal labeled examples, including crater identification, terrain mapping, and the segmentation of Irregular Mare Patches.

A primary application of the model is the detection of potential water ice deposits near the lunar poles. In internal testing, the model reportedly reduced errors in detecting ice areas by 23 percent compared to Microsoft’s SwinV2-B system.

The model was trained using SomBench, a large multimodal lunar corpus containing nearly 2 million tile bundles. This dataset integrates 17 years of observations from multiple missions, including NASA’s Lunar Reconnaissance Orbiter, GRAIL, Lunar Prospector, and JAXA’s Kaguya/SELENE probe, providing over 30 spatially aligned data layers.

## Timeline

### 2026-10-04: NASA and IBM release open-source lunar foundation model

NASA and IBM Research have launched an open-source lunar foundation model trained on 17 years of orbital data to assist in crater detection and predicting lunar ice deposits.

2 sources. https://clstr.news/cluster/nasa-and-ibm-release-open-source-lunar-foundation-model

### 2026-09-13: NASA and IBM release open-source AI model for lunar exploration

NASA and IBM have released the NASA-IBM Lunar Foundation Model, an open-source AI designed to map the Moon, identify craters, and locate vital water ice deposits to support future space exploration.

10 sources. https://clstr.news/cluster/nasa-and-ibm-release-open-source-ai-for-lunar-exploration

---
Cite as: NASA-IBM lunar exploration AI model release. CLSTR, https://clstr.news/situations/nasa-ibm-lunar-exploration-ai-model-release