WeatherNext: AI model achieves breakthrough in forecasting cyclones
Google DeepMind open-sourced WeatherNext code and weights, reporting a full day of extra cyclone lead time and 1,000-member ensembles.
- Ensemble size scaled to 1,000 members, capturing rapid intensification events like Hurricane Melissa in 2025.
- WeatherNext Cyclones runs at 28x28km resolution, 100x coarser than traditional models.
- WeatherNext 2-mini operates on a single TPU in a free public Colab notebook.
- Weather Lab refreshed with a new interface and expanded to global weather forecasts alongside cyclone tracks.
Meteorological agencies, researchers, and nonprofits can now build on open-source WeatherNext models to improve cyclone forecasts and preparedness.
Sources
Read this as text
Back to the AI news