{"version":1,"type":"story","url":"https://digestai.news/story/nasa-and-ibm-release-open-source-lunar-foundation-ai-model","json":"https://digestai.news/story/nasa-and-ibm-release-open-source-lunar-foundation-ai-model.json","markdown":"https://digestai.news/story/nasa-and-ibm-release-open-source-lunar-foundation-ai-model.md","slug":"nasa-and-ibm-release-open-source-lunar-foundation-ai-model","headline":"NASA and IBM release open-source lunar foundation AI model","summary":"NASA and IBM Research announced the open‑source NASA‑IBM Lunar Foundation Model, a multimodal AI system pretrained from scratch on SomBench, a lunar dataset containing nearly two million co‑registered data bundles across 11 modalities and two spatial scales. The model integrates imagery, topography, illumination geometry, thermophysical properties, mineralogy, radar, gravity and other lunar observations.\n\nThe model was evaluated on three downstream benchmarks—regional and meter‑scale crater detection, segmentation of irregular mare patches, and regression of lunar polar ice prospectivity—and matched or outperformed ImageNet‑pretrained baselines and a comparable model without lunar pretraining. USRA’s Dr. Rachel Slank contributed planetary‑science expertise, helped build the high‑resolution LROC NAC crater benchmark with more than 49,000 manually identified craters, and reviewed the study. The pretrained model, fine‑tuning code and benchmark datasets are released openly on Hugging Face for the planetary‑science and AI communities.","keyPoints":["Model pretrained on SomBench, a multimodal lunar dataset of nearly two million bundles across 11 modalities","Evaluated on crater detection, irregular mare patch segmentation, and polar ice prospectivity, matching or beating ImageNet‑pretrained baselines","USRA’s Dr. Rachel Slank helped create benchmarks, identifying over 49,000 lunar craters; model released on Hugging Face"],"whyItMatters":"Open‑source domain‑specific foundation model shows label‑efficient lunar analysis and speeds scientific discovery, highlighting value of specialized pretraining.","category":{"slug":"models","name":"Generative AI & Models","url":"https://digestai.news/category/models"},"entities":{"companies":["NASA","IBM Research","Universities Space Research Association"],"models":["NASA-IBM Lunar Foundation Model","SomBench"],"people":["Dr. Rachel Slank"]},"firstPublishedAt":"2026-09-19T04:44:35Z","updatedAt":"2026-09-19T04:44:35Z","sourceCount":1,"hasPrimarySource":false,"sources":[{"outlet":"newsroom.usra.edu","title":"NASA-IBM Lunar Foundation open-Source Geospatial AI Model","url":"https://newsroom.usra.edu/usra-contributes-planetary-science-expertise-to-nasa-ibm-lunar-foundation-model","publishedAt":"2026-09-19T04:44:35Z","type":"press","primary":false,"lead":true}],"sourceNotes":null,"discussions":[{"site":"Hacker News","url":"https://news.ycombinator.com/item?id=49763379","points":42}],"thread":null,"cite":{"text":"Digest AI, \"NASA and IBM release open-source lunar foundation AI model\", 19 September 2026, https://digestai.news/story/nasa-and-ibm-release-open-source-lunar-foundation-ai-model","publisher":"Digest AI","title":"NASA and IBM release open-source lunar foundation AI model","datePublished":"2026-09-19T04:44:35Z","url":"https://digestai.news/story/nasa-and-ibm-release-open-source-lunar-foundation-ai-model"},"generatedBy":"Written by Digest AI's editorial model from the linked sources; the sources are the record.","license":"Headlines, digests and key points are written by Digest AI and may be quoted with a link to the story page. Linked articles belong to their publishers. Terms: https://digestai.news/terms#reuse"}