Otter Weather: Skillful and Computationally Efficient Medium-Range Weather Forecasting
Paper โข 2606.26421 โข Published
An Otter weather forecasting model,
from experiment rft_final_2026-03-06_22-07-wild-lynx at step 18977, trained with code commit
61a9bc0a9c1860321e3e624e3ea8fe55909b243c.
gs://weatherbench2/datasets/era5/1959-2023_01_10-6h-240x121_equiangular_with_poles_conservative.zarr.Forecasting needs a CUDA GPU.
git clone https://github.com/cambridge-mlg/otter && cd otter && uv sync
uv run python -m otter.hub forecast jonasscholz/otter-xl \
--init_time 2020-07-01T00 --output forecast.zarr
The forecast is a zarr store following the WeatherBench2 conventions. The
config is instantiated with Hydra, which imports and calls the code it names:
pin the revision with --revision <commit hash>.
The weights are released under the MIT License, as is the code; see
LICENSE.
@article{diaconu2026otter,
title={Otter Weather: Skillful and Computationally Efficient Medium-Range Weather Forecasting},
author={Diaconu, Cristiana and Scholz, Jonas and Shysheya, Aliaksandra and Markou, Stratis and Mukhopadhyay, Payel and Cranmer, Miles and Turner, Richard E},
journal={arXiv preprint arXiv:2606.26421},
year={2026}
}