WeatherNext 3 beat leading AI and government forecast models and will power Google Search, Maps, and Gemini.
Google DeepMind and Google Research released WeatherNext 3, an AI weather model that forecasts hourly at five-kilometer resolution. The model will feed weather data into Google Search, Google Maps, and Gemini, and will be available on Google Cloud platforms for external users.
Traditional weather forecasting uses supercomputers that are accurate but expensive and slow. After the European Center for Medium-Range Weather Forecasting released decades of data in 2018, deep learning researchers began building models that predict far more quickly and with comparable accuracy.
For builders and operators, WeatherNext 3 fixes the biggest issues with AI forecasts. The model predicts at five-kilometer resolution, improves rain accuracy by 60 percent over WeatherNext 2, and produces forecasts hourly instead of every six hours. Teams can access the model on Google Cloud and integrate its output into their own applications.
WeatherNext 3 topped Operational WeatherBench, outperforming other AI and government forecast models. Google will start using the model to power weather information in Search, Google Maps, and Gemini. The model uses 2.4 times more parameters than its predecessor, giving it the capacity for hourly five-kilometer forecasts.
What matters
- Google DeepMind released WeatherNext 3, an AI model that forecasts hourly at 5km resolution.
- For builders, the model offers more accurate rain forecasts and hourly updates than AI predecessors.
- Watch for WeatherNext 3 to power weather data inside Google Search, Google Maps, and Gemini.
Why it matters
Watch for WeatherNext 3 to power weather data inside Google Search, Google Maps, and Gemini.
This GenAI News article was prepared in original wording using reporting and materials published by TechCrunch AI. Source reference: https://techcrunch.com/2026/09/03/googles-latest-ai-weather-model-gives-you-no-excuse-to-forget-your-umbrella/.
Drafted by the GenAI News review pipeline.