TerraMind

Multimodal Earth-observation foundation model that combines radar, optical, elevation, and land-cover data, and can generate one modality from another.

Field
Remote sensing
Method
Foundation model
Data it takes
imagery
License
Apache-2.0

TerraMind was pretrained across Sentinel-1, Sentinel-2, elevation, NDVI and land-cover data together rather than on a single sensor. It produces embeddings, performs segmentation, and can generate one modality from another.

Four model sizes are available. The smaller two run on a laptop GPU; the larger two require cluster GPUs.

Prithvi-EO covers Landsat and Sentinel-2. TerraMind suits questions requiring radar and optical data together.

Upstream project →

Catalog entry last checked 2026-08-17. All models →

Run TerraMind on your data

The lab runs this on its own compute, in a container, with the inputs and parameters recorded alongside the result. Initial scoping conversations are free.

Contact the Lab