Google puts WeatherNext 3 into Search, Gemini and Maps

Google DeepMind and Google Research introduced WeatherNext 3, a global weather AI model that is moving directly into Google products and cloud workflows. The model now powers weather experiences in Search, the Gemini app, Maps, Google Maps Platform Weather API, Google Earth Engine and Google Cloud data access.

The main technical shift is fresher data and finer detail. WeatherNext 3 ingests live geostationary satellite observations, produces new forecasts every hour and can model key surface variables such as temperature and moisture at up to 5-kilometer resolution. Google says that is roughly five times sharper than WeatherNext 2, which used a 25-kilometer grid and six-hour increments.

Google also claims a major step forward in precipitation forecasting, with evaluations showing up to 60% better probabilistic scores against satellite rainfall data in some medium-range comparisons. For users, the company says longer-range precipitation forecasts in Google products can become up to 50% more accurate, especially in regions where traditional forecasting has been weaker.

The broader point is distribution. WeatherNext 3 is not just another research benchmark: Google is putting an AI forecast model into products people already use, while also making high-resolution data available for developers, researchers, energy planners and businesses through BigQuery, Earth Engine and Cloud Storage. Official public warnings still belong to meteorological agencies, but AI forecasting is becoming part of everyday infrastructure.