Google’s WeatherNext 3 is now rolling out as the company’s most advanced global weather AI model, delivering immediate upgrades to Search, Maps, and the Gemini app. Developed by Google DeepMind and Google Research, the system promises significantly sharper and more localized predictions than the versions that came before it.
Earlier AI weather systems, including WeatherNext 2 released in November 2025, were trained on data from numerical weather prediction (NWP) models. Those models rely on complex, supercomputer-driven physics simulations that carry a six-hour data lag. That delay can introduce bias when forecasting fast-changing conditions such as rainfall and surface temperature.
How WeatherNext 3 Works
Rather than depending solely on NWP data, WeatherNext 3 learns from a mosaic of live, global geostationary satellite observations. This real-time approach produces a continuously updating view of the atmosphere, enabling more timely and localized forecasts. The model can generate hourly predictions at several spatial resolutions.
Temperature and moisture are resolved at 5-kilometer resolution, surface variables at 10 kilometers, and wind speed at 25 kilometers. Compared with WeatherNext 2, which operated on a 25-kilometer grid in six-hour increments, the new version delivers a global weather picture that is roughly five times sharper. Because storms, fronts, and precipitation systems can develop suddenly, the faster update cycle and higher resolution are designed to provide earlier, more detailed insights.
A Major Jump in Precipitation Accuracy
Google credits much of the improvement to two high-quality data sources used in training: NASA’s satellite-based Integrated Multi-satellite Retrievals for GPM (IMERG) and Google’s own global precipitation reanalysis based on satellite radar. In medium-range global forecasts, evaluations against baselines showed a Continuous Ranked Probability Score (CRPS) improvement of up to 60% against IMERG, 30% for MRMS, and 10% against rain gauge measurements for early lead times.
The company says the model is especially important for regions across Latin America, Africa, and Asia-Pacific that have historically been underserved by high-resolution forecasting due to the steep supercomputing costs of traditional regional models. WeatherNext 3 is intended to bring localized, high-fidelity forecasting to billions of people and local businesses in those areas.
Renewable Energy and Consumer Rollout
WeatherNext 3 also introduces predictions built specifically for renewable energy production. The model forecasts 100-meter wind speeds, roughly at turbine height, for more precise wind-energy output estimates. It also provides high-resolution cloud cover and solar radiation data to help solar farms gauge how much sunlight will reach the ground.
On the consumer side, the model is rolling out to Google Search, Maps, and the Gemini app, where Google says users can expect dramatically improved longer-term forecasts. When planning a day or more ahead, people will see up to 50% more accurate precipitation forecasts, with
Source
Image: 9to5google.com
