Otter Weather: Skillful and Computationally Efficient Medium-Range Weather Forecasting
Cristiana Diaconu, Jonas Scholz, Aliaksandra Shysheya, Stratis Markou, Payel Mukhopadhyay, Miles Cranmer, Richard E. Turner · 2026
Read Otter Weather: Skillful and Computationally Efficient Medium-Range Weather Forecasting by Cristiana Diaconu, Jonas Scholz, Aliaksandra Shysheya, Stratis Markou, Payel Mukhopadhyay, Miles Cranmer, Richard E. Turner (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2606-26421.txt.
Description
Cristiana Diaconu and six colleagues built Otter Weather. It is an artificial intelligence model for medium-range weather forecasts. Large AI forecast models beat traditional physics-based prediction on many scores. Yet their huge training budgets shut out smaller research groups. Otter was designed to open that work up at far lower cost.
Tests used ERA5, a standard record of past weather. Otter beat the best traditional baseline by about ten percent at one day ahead. Training took under three and a half days on a top graphics chip. A larger version, Otter-XL, also improved a standard probabilistic score by nearly ten percent. That score grades how well predicted chances match outcomes. Otter even handled an unrelated sound-scattering physics task well, with no retraining at all.
Catalog record
| CREATOR | Cristiana Diaconu, Jonas Scholz, Aliaksandra Shysheya, Stratis Markou, Payel Mukhopadhyay, Miles Cranmer, Richard E. Turner |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 83 KB |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2606.26421