NASA and IBM Research are releasing a joint AI foundation model aimed at weather prediction and climate analysis, according to NASA's open science program page. The model will be publicly available, consistent with NASA's open science mandate.
The release extends IBM Research's Prithvi model series, which has targeted geoscience applications. NASA's involvement adds access to its satellite and Earth observation datasets, which underpin the model's training scope — though specific dataset sources and model architecture details have not been disclosed in the available signal.
The project sits within a broader institutional push toward domain-specific foundation models in Earth science, where general-purpose LLMs have shown limited utility against specialized observational data formats like hyperspectral imagery and gridded climate outputs.
For climate tech developers and geospatial AI operators, a NASA-credentialed, openly licensed foundation model reduces the barrier to fine-tuning on sector-specific tasks — eliminating the need to source or negotiate access to comparable pretraining data independently.