Model reference · open weights
OlmoEarth-v1_2 is an open-weight embedding model from allenai. OlmoEarth-v1_2-Base (BF16) weighs 1.0 GB; the smallest configuration that runs it is RTX 3060 12 GB.
Summary of the allenai/OlmoEarth-v1_2-Base model card, 2026-10-01
What it is
| Released by | allenai |
|---|---|
| Released | 2026-06-25 |
| VRAM | 1.0 GB for the weights |
What it runs on
| Card | Runs | Memory |
|---|---|---|
| RTX 3060 12 GB | fits | 11.6 GB |
| RTX 4060 Ti 16 GB | fits | 15.4 GB |
| RTX 3090 24 GB | fits | 23.4 GB |
| RTX 4090 24 GB | fits | 23.4 GB |
| RTX 5090 32 GB | fits | 31.0 GB |
| L40S 48 GB | fits | 44.0 GB |
| A100 80 GB | fits | 78.2 GB |
| H100 80 GB | fits | 78.1 GB |
| RTX PRO 6000 Blackwell 96 GB | fits | 93.8 GB |
| DGX Spark (GB10) 128 GB unified | fits | 107 GB |
| H200 141 GB | fits | 138 GB |
| B200 180 GB | fits | 176 GB |
From the model card
OlmoEarth-v1_1-Base is a ViT-Base (114M parameters) foundation model for remote sensing tasks over Sentinel-1, Sentinel-2, and Landsat images or image time series.
This model is presented in the paper OlmoEarth v1.2: A more efficient family of OlmoEarth models.
Please see our GitHub and paper for more information about OlmoEarth.
If you are interested in computing embeddings with OlmoEarth, or fine-tuning OlmoEarth for a downstream task, see our tutorials:
The OlmoEarth model is deployed in the OlmoEarth Platform, a solution that enables any organization to leverage our models and accelerate their mission, without requiring deep AI or engineering expertise. Sign up to get updates and partner access.
This model is licensed under the OlmoEarth Artifact License. It is intended for use in accordance with Ai2's Responsible Use Guidelines.
Quoted from the model card on Hugging Face. The full card is behind the Hugging Face link above.
How it works