DeepMind's WeatherNext model achieves breakthrough forecasting cyclones

DeepMind's WeatherNext model has made a breakthrough in forecasting cyclones, providing an extra day of warning compared to previous models. This model can accurately predict a cyclone's track, intensity, and wind structure, improving forecasters' predictive accuracy by an entire day. This improvement is equivalent to a decade's worth of meteorological progress.

The collaborative effort between AI researchers and engineers at Google DeepMind and Google Research, along with expert forecasters at various institutions, resulted in this significant advancement. During the 2025 hurricane season, WeatherNext helped the National Hurricane Center (NHC) make a historic forecast for Hurricane Melissa, enabling them to issue an advance warning that gave local teams critical time to prepare.

In response to this success, DeepMind is now open sourcing the WeatherNext 2 and WeatherNext Cyclones models, making them available for use by researchers, forecasters, and organizations worldwide. This open-source release aims to empower the research community and enhance the impact of AI in building more resilient communities by providing tools for preparing for natural disasters, supporting renewable energy growth, and anticipating extreme weather events.