Google has launched WeatherNext 3, calling it its most advanced and accurate global weather AI model yet. Google DeepMind and Google Research unveiled the model on September 3, 2026. It now powers weather experiences across Search, Gemini, Maps, and Google Cloud.
The model’s biggest leap involves how it learns. Most AI weather models train on numerical weather prediction data. However, those physics simulations carry a six-hour data lag. That delay creates biases for fast-changing variables like rain and surface temperature. WeatherNext 3 instead ingests live geostationary satellite data directly.This approach unlocks major improvements in speed and detail. The model generates a fresh forecast every single hour. Each one draws on the most recent satellite observations available. It visualizes temperature and moisture at five-kilometer resolution. Overall, this delivers a global weather picture roughly five times sharper than WeatherNext 2.
The accuracy gains prove especially strong for precipitation. Global models notoriously struggle to predict rain and snow well. WeatherNext 3 trains on high-quality data from NASA satellites and radar-based reanalysis. As a result, medium-range forecasts show accuracy improvements of up to 60% against certain baselines.The model carries particular significance for underserved regions.
Historically, high-resolution forecasting required immense supercomputing costs. This left much of Asia, Africa, and Latin America poorly served. Because WeatherNext 3 trains on sparse weather station data, it delivers localized forecasts accounting for topography. This brings high-fidelity forecasting to billions of people previously overlooked.
Accurate, localized rainfall prediction remains a persistent challenge across South Asia. Since the model improves forecasts precisely where they have been least reliable, it could aid disaster preparedness meaningfully. However, Google stressed that official warnings should still come from national meteorological agencies.
The model also introduces variables tailored for renewable energy. It forecasts wind speeds at turbine height for precise energy output. Additionally, it predicts cloud cover and solar radiation for solar farms.
When planning a day or more ahead, users will see up to 50% more accurate precipitation forecasts. Developers and researchers can also access the underlying data through Google Cloud, BigQuery, and Earth Engine for their own projects.
