Free Satellite AI Now Lets Anyone Map Earth's Changes

A new AI model called OlmoEarth v1.3 has been released, and it can analyze satellite images of the entire planet to create detailed maps showing what's on the ground. The model turns complex satellite data into simple numerical representations called embeddings, which can then be used to track changes over time or find similar areas. Best of all, it's completely free and open-source, and comes with a browser-based tool that lets anyone explore these maps without needing any technical expertise.

This matters because it puts powerful environmental monitoring into the hands of everyone, not just scientists with big budgets. For example, you can use it to spot illegal gold mining in the Amazon, map small farms in Kenya, or assess fire damage in Indonesia. The embeddings are precomputed for the whole world, so you don't need to download massive amounts of data or have special hardware. You can just use the online explorer to see changes over time, like how a mining area expanded between 2017 and 2020 and then shifted elsewhere by 2025.

The model is also a big step forward in efficiency. Training it took only 18% of the computing time compared to the first version, and it produced half the carbon emissions. Despite being cheaper to run, it outperforms other similar models on most tests, especially in places it hasn't seen before. All the details, including the model's code and data, are publicly available, so anyone can verify or build upon it.

The catch? While the tool is free and easy to use, interpreting the results still requires some basic understanding of what the colors and patterns mean. But with the interactive explorer and documentation, even non-experts can start exploring. This release marks a significant shift toward democratizing access to planetary-scale environmental data, making it possible for journalists, activists, and curious citizens to monitor the health of our planet like never before.