Hugging Face Releases OlmoEarth v1.3, a 128‑dimensional Satellite Embedding Model for the Entire Planet

Hugging Face has launched OlmoEarth v1.3, a vision‑transformer based model that compresses a full year of Sentinel‑1, Sentinel‑2 and Landsat imagery into a 128‑dimensional vector for every 10‑meter pixel on Earth. The release includes pre‑computed annual embeddings for the whole planet, a 64‑dimensional subset, and a browser‑based Embeddings Explorer that lets users map, compare, and classify land‑cover without GPU access. In 27 evaluations across six continents, OlmoEarth’s embeddings outperformed AlphaEarth Foundations and Tessera v2 on 20 of 27 tasks, with the 64‑dimensional version still leading on 18 tasks.