Google DeepMind releases WeatherNext 3, an updated global weather AI model intended to improve precipitation forecasting. The company says the system can produce more accurate precipitation predictions than prior approaches, including forecasts a day or more in advance.
The outlets describe the model as using near-real-time inputs, including satellite data, to generate weather forecasts at frequent intervals. One source notes that WeatherNext 3 produces updates hourly and that it models surface variables such as temperature and humidity, alongside precipitation outcomes.
While both accounts focus on the model’s capabilities and claimed accuracy gains, they differ slightly in emphasis: one highlights the magnitude of precipitation accuracy improvement and the lead time, while the other stresses how the model works operationally—pulling in real-time satellite observations and refreshing forecasts on an hourly cadence. Both characterize WeatherNext 3 as DeepMind’s most advanced global weather model to date.
