Google DeepMind has introduced WeatherNext Cyclones, or WN-C, an AI system for tropical cyclone forecasting that sees about one day further into the future than leading operational models. The gain roughly matches the progress traditional weather models made over a decade.
The model was built with the National Hurricane Center, the Cooperative Institute for Research in the Atmosphere, and the UK Met Office. Forecasts have been running live on Google's Weather Lab since June 2025. During Hurricane Melissa, which made landfall in Jamaica in 2025, the model helped the NHC predict rapid intensification in time, according to DeepMind. That is when a storm gains at least 30 knots of wind speed within 24 hours.
Solving a decades-old tradeoff
Cyclone forecasting has long faced a tradeoff. Global models like ECMWF's ensemble system (ENS) are strong on track prediction but too coarse for intensity. Specialized regional models like NOAA's Hurricane Analysis and Forecast System (HAFS) deliver more precise intensity readings but lose accuracy on the track. WN-C handles both in a single system, according to a paper published in Nature.
For a five-day forecast, the estimated storm center position is off by an average of 230 kilometers, compared to 370 for ENS and 335 for DeepMind's predecessor GenCast. On three-day intensity forecasts, WN-C is 3.75 knots more accurate than HAFS. It also scores more than twice as well as ENS and GenCast on probabilistic storm intensity forecasts across many lead times.
Coarse data, sharper forecasts
WN-C works with a data grid where each point covers about 28 kilometers, roughly a hundred times coarser than specialized regional models. Even a compact variant at 111 kilometers per grid point stays competitive. "High resolution is not a strict prerequisite for state-of-the-art intensity forecasting," the authors write, noting that how the model extracts accurate intensity from coarse data remains an open research question.
Faster than diffusion
WN-C uses Functional Generative Networks (FGN) instead of the diffusion approach used by GenCast. Diffusion needs dozens of passes through the network per forecast step; FGN gets by with a single pass, making it eight times faster. A 15-day forecast runs in under a minute on one of Google's AI chips, so DeepMind scaled parallel runs from 50 to 1,000 per storm, which helps capture rare extreme events. The Critical Success Index, which weighs correct warnings against false alarms and missed events, jumps from below 0.3 to 0.5.
A complement, not a replacement
DeepMind says traditional numerical models still matter. The NHC typically combines several physics-based models into a weighted consensus for official forecasts. In a simulated addition to those consensus models, WN-C improves track forecasts by an average of 28 percent and intensity by about 6 percent. The smaller intensity gain shows traditional models still contribute heavily there.
DeepMind released the code and weights for both WeatherNext 2 and WeatherNext Cyclones on GitHub, and the mini variant runs on a single TPU in a free Colab notebook. The authors still point users to national weather services for official warnings: WN-C is meant to support forecasters, not replace them.
