Google's WeatherNext 3 AI model makes hourly forecasts in much finer detail

On September 3, 2026, Google DeepMind and Google Research introduced WeatherNext 3, an AI model that forecasts the weather for the whole planet. It makes a forecast for every hour of the day, while its predecessor, WeatherNext 2, worked in six-hour steps. For conditions near the ground, such as temperature and humidity, it works at a resolution of about 5 kilometers, which Google says gives a picture roughly five times sharper than before. The model learns from live satellite data, and Google says people can expect precipitation (rain and snow) forecasts up to 50% more accurate. Google also says live tests by an independent company, Brightband, found it to be the most accurate global weather model so far. WeatherNext 3 already powers weather information in Google Search, Google Maps and the Gemini app, and researchers can reach its data through Google Cloud tools. TechCrunch noted that it still relies on datasets from national weather services. Why it matters: Weather forecasting shows how AI can learn patterns from huge amounts of real-world data, and more accurate rain forecasts are useful to almost everyone. Source: Google — https://blog.google/innovation-and-ai/models-and-research/google-deepmind/introducing-weathernext-3/ Source: TechCrunch — https://techcrunch.com/2026/09/03/googles-latest-ai-weather-model-gives-you-no-excuse-to-forget-your-umbrella/