FourCastNet 3

Produces global weather forecasts at 0.25-degree resolution, generating an ensemble rather than a single trajectory.

Field
Weather & climate
Method
Deep learning
Data it takes
gridded-fields
License
Apache-2.0

FourCastNet 3 forecasts global atmospheric state on a 0.25-degree grid. It is probabilistic by construction: each run produces an ensemble, so forecast spread comes out of the model rather than being assembled from separate runs.

Machine-learning forecast models run orders of magnitude faster than numerical weather prediction at comparable skill for many variables, which makes large ensembles and rapid re-forecasting practical.

Code and weights are both Apache-2.0.

Upstream project →

Catalog entry last checked 2026-08-19. All models →

Run FourCastNet 3 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