IBM and NASA open-source Lunar Foundation Model

10 Sep 2026: IBM Research and NASA Science both announced the open-source NASA-IBM Lunar Foundation Model — a multimodal, multi-resolution lunar remote-sensing foundation model trained primarily on Lunar Reconnaissance Orbiter data plus other mission archives. Hugging Face + TerraTorch. Benchmark % gains stay IBM/NASA’s claims.

10 Sep 2026: IBM Research published “Introducing IBM and NASA’s new foundation model for the Moon,” and NASA Science published “NASA, IBM Launch AI Foundation Model for Lunar Science” (Rachel Wyatt). Both name the open-source NASA-IBM Lunar Foundation Model. Those dated agency pages are the filing event.

IBM: the model is the first AI model to integrate lunar observations captured across modalities, viewing angles, and spatial scales, and it sits in the lineage of earlier IBM/NASA Earth and Sun models Prithvi EO and Surya. NASA’s wording is more cautious: among the first open-source AI models built specifically for lunar science. Those are the agencies’ framings — not an independent “first” audit.

NASA: trained primarily on Lunar Reconnaissance Orbiter (LRO) data — roughly 2 million image tiles, comprising more than 1 million high-resolution camera images at 1-meter resolution and nearly 964,000 multispectral images at 100-meter resolution. Also trained on imagery and terrain data from NASA’s GRAIL, NASA’s Lunar Prospector, and JAXA’s Selenological and Engineering Explorer (SELENE).

IBM: the architecture is a version of TerraMind, the Earth-observation model developed by IBM and ESA. Fine-tuning used lightweight LoRA adapters that left 90% of the base model’s weights frozen. NASA: weights are hosted on Hugging Face, the complete codebase is on GitHub, and the model is integrated into the open-source TerraTorch toolkit.

Priority tasks on both pages: mapping smaller uncatalogued craters; investigating volcanic history via irregular mare patches (IMPs); and prospecting polar craters for ice. IBM lists those as NASA’s initial priorities. NASA: the pre-trained model can be adapted to those tasks with only small amounts of labeled data.

IBM’s published comparisons versus task-specific Swin / SwinV2 models: a 22% cut in error rate identifying whether a dark polar crater might contain ice, versus a SwinV2 transformer trained for ice prospecting; crater detection at 1 meter-per-pixel matched a Swin model trained for crater detection; at 100 meters-per-pixel the NASA-IBM model outperformed that same model by nearly 19% using half the training data; on mapping irregular mare patch extents it outperformed a task-specific Swin by 3%. NASA’s page is more cautious: the model matched or exceeded several strong baselines, with comparable results on crater mapping and IMP segmentation and a clear advantage on estimating polar ice stability. Those percentages and rankings are IBM’s / NASA’s benches — this desk did not rerun them.

This is a mapping / analysis tool drop, not a lander or Artemis schedule announcement. IBM’s page puts the model in the context of Artemis-era exploration and polar-ice prospecting; it does not announce a landing date, lander, or certified landing site.

Open multimodal lunar foundation model with dual IBM+NASA primaries — Artemis-era mapping tooling, not a lander announcement.