Google launched WeatherNext 2, an AI-powered weather forecasting model developed by Google DeepMind and Google Research today. The announcement coincided with the reveal of Gemini 3 Pro and Antigravity.
WeatherNext 2 uses a Functional Generative Network (FGN) architecture that simulates hundreds of possible weather scenarios from a single input. This approach delivers forecasts up to eight times faster than previous models. It shows superior accuracy across 99.9% of tested variables, including temperature, wind speed, and humidity.
The model is now available in Google Search, where users see improved forecasts directly in search results with hour-by-hour details. Gemini supports conversational weather queries. Pixel Weather provides precise updates on compatible devices.
Developers and enterprises can access WeatherNext 2 datasets through Google platforms. Earth Engine enables geospatial data integration for environmental analysis. BigQuery handles large-scale data querying and analytics. Early-access APIs in Vertex AI support programmatic integration into custom applications.
These enhancements aid planning for commutes, travel itineraries, agricultural schedules, energy grid management, and emergency response protocols. With extreme weather events increasing, the model’s short-range and long-range predictions improve safety measures and preparedness.
Google Search offers hour-by-hour details for immediate planning. Google Maps integration includes upgraded weather layers, real-time overlays, clearer routing adjusted for weather conditions, precise storm tracking, and alerts for travel or outdoor activities.
Google confirmed Google Maps as the next major product to adopt WeatherNext 2. This update brings advanced weather layers, real-time overlays, and targeted alerts within navigation and travel tools. Power users and businesses use the improved datasets to optimize deliveries, event scheduling, and outdoor operations.
Google plans personalized insights from the model, such as predictive rain windows tailored to individual commutes and automated task suggestions based on forecast changes.
