The model weights, pretraining data, and code are available alongside the data archive.
OlmoEarth v1.3 processes yearly Sentinel-1 radar plus Sentinel-2 and Landsat multispectral time series. It produces a 128-dimensional vector for each 10-meter pixel; the first 64 dimensions can also be used as a smaller embedding. Ai2 says users can explore changes between years, find similar locations, label points, train a small classifier, and export GeoTIFF results in the Explorer without an account, setup, or GPU.
Ai2 reports that its 128-dimensional embeddings ranked first on 20 of 27 evaluations against AlphaEarth Foundations and Tessera v2, while the 64-dimensional version led 18. It estimates the final pretraining run would cost about $2,720 in rented H100 time. These are Ai2's reported benchmarks and cost estimate; the source does not independently establish performance for every geography or downstream task.