GenCast
GenCast is a diffusion-based ensemble weather model that makes a 15-day probabilistic forecast in 8 minutes on one TPU and beats ECMWF ENS on 97.2% of 1,320 targets.
Generates 50+ weather trajectories at 0.25 degrees, trained on 40 years of ERA5. Better at extreme events, cyclone tracks and wind-power forecasting. Code and weights released.
- Date
- Wednesday, 4 December 2024
- Lab
- Google DeepMind
- Kind
- model
- Access
- open weights
Figures
| Measure | Value | Measured by |
|---|---|---|
| Targets beating ECMWF ENS | 97.2% of 1,320 99.8% beyond 36-hour lead time | company |
Published in Nature 2024-12-04.
Sources
This record was checked against its sources on 6 October 2026. How we check