Multi-Modal Foundation · allenai
OlmoEarth v1 Base
A remote sensing foundation model for analysing Sentinel-1, Sentinel-2, and Landsat imagery and time series.
What OlmoEarth v1 Base is
OlmoEarth v1 Base is an Earth-observation foundation model developed by the Allen Institute for AI. Designed for earth science, climate research, and conservation studies, it provides pre-trained visual representations for remote sensing workflows.
The model balances model capacity with inference efficiency across Earth-observation workloads. It processes individual scenes as well as temporal sequences captured by major satellite programmes.
How it works
The model is built on a Vision Transformer Base (ViT-Base) architecture comprising 89 million parameters. It accepts inputs from Sentinel-1 radar, Sentinel-2 optical, and Landsat satellite sensors in the form of individual images or multi-temporal image series.
Key facts
- Architecture
- ViT-Base
- Parameters
- 89 million
- Supported sensors
- Sentinel-1, Sentinel-2, Landsat
- Input modalities
- Single images and image time series
What you can use it for
- Satellite image segmentation
- The model can be fine-tuned to segment geographical features and land-cover classes in remote sensing imagery.
- Embedding extraction
- Researchers can compute dense feature representations from satellite inputs for downstream analytical workflows.
- Time-series remote sensing
- The architecture handles temporal sequences of Earth-observation images to analyse changes over time across Landsat and Sentinel data.
Run it here
We list this model but cannot run it yet: no pipeline_tag — needs a human.
Everything on this page still applies. Running it needs a loader for this repository, not a bigger machine.
Strengths and limits
Good at
- Processes data across multiple distinct satellite platforms, including Sentinel-1, Sentinel-2, and Landsat.
- Natively supports both single-date satellite acquisitions and multi-image temporal series.
- Offers a 89-million-parameter footprint designed to balance representational capability and inference compute requirements.
Watch out for
- Source data support is specifically focused on Sentinel-1, Sentinel-2, and Landsat observations.
- Applying the model to specialised downstream tasks requires supplementary fine-tuning or embedding generation pipelines.
Inputs and outputs
You provide
Sequence or text
A protein or nucleotide sequence, or free text — whichever this model was trained on.
You get back
Mean-pooled representation, one vector per input.
Licence
Check the licence before commercial use · See model card
The See model card licence carries terms worth reading before you depend on it.
Questions
Can I use OlmoEarth v1 Base commercially?
Check the licence before commercial use. The See model card licence carries terms worth reading before you depend on it.
What sensor types can this model process?
The model accepts imagery from Sentinel-1 synthetic aperture radar as well as Sentinel-2 and Landsat optical sensors.
Can this model handle satellite time series?
Yes, the architecture is designed to handle both individual static images and image time series.
How large is OlmoEarth v1 Base?
The model contains 89 million parameters organised in a ViT-Base architecture.
Written from OlmoEarth v1 Base's own documentation (model card). Statements citing a figure that does not appear there are removed before publication. Licence and running details come from our registry.