Google has launched WeatherNext 3, an updated AI weather forecasting model that the company says produces significantly sharper and more frequent predictions than its predecessor, particularly for rain and snowfall. Unlike traditional forecasting, which relies on supercomputers simulating atmospheric physics, WeatherNext 3 incorporates live satellite data alongside historical patterns, allowing it to generate hourly forecasts and fill in gaps in regions with fewer ground-based rain gauges, mainly outside the US and Europe.

The new model offers a global picture five times sharper than its predecessor, WeatherNext 2, which produced forecasts every six hours on a 25-kilometre grid; WeatherNext 3 can visualise variables such as temperature and moisture at up to 5-kilometre resolution. Google says this yields precipitation forecasts up to 50% more accurate a day or more in advance, and the model also predicts wind speeds at turbine height to support renewable energy planning. WeatherNext 3 is now built into Search, Maps and Gemini, and Google has partnered with the US National Hurricane Center and Asian agencies, though AI forecasts still work alongside, not instead of, physics-based models.