Back to products
WeatherNext By Google

WeatherNext By Google

Google's most advanced weather forecasting AI technology

Website deepmind.google
Overview

What it is

We’re putting our most advanced AI for weather forecasting into more people’s hands. Scientists can now access Google’s WeatherNext models as well as real-time and historical data to accelerate research, improve how we respond to disasters and much more.

Intent

I need it when

Improve forecast accuracy for low-probability but catastrophic weather events

WeatherNext 2 analyzes higher numbers of possible scenarios per forecast due to 8x speed improvement, making it significantly better at predicting rare but dangerous events like rapid hurricane intensification that traditional models struggle to forecast.

Generate rapid, accurate weather forecasts for operational decision-making by meteorological agencies

WeatherNext 2 is 8x faster than previous models, generates six-hour forecasts four times daily, and forecasts wind speed, direction, precipitation, and pressure with high accuracy. This enables national weather services to support emergency response and resource mobilization.

Integrate AI weather forecasting into enterprise systems and geospatial analysis platforms

WeatherNext 2 is available via Gemini Enterprise Agent Platform, BigQuery, and Google Earth Engine, allowing businesses and research organizations to customize models and create forecasts for downstream applications and advanced geospatial analysis.

Predict extreme weather events and tropical cyclone intensity with high accuracy days in advance

WeatherNext 2 predicts both track and intensity of hurricanes and cyclones up to 15 days in advance with ensemble forecasting of 50 scenarios. It achieved 80% confidence predicting Hurricane Melissa as Category 5 five days before landfall, enabling meteorological agencies to issue early warnings and coordinate evacuations.

Drop

Not a fit when

  • User needs real-time weather alerts for immediate personal use; WeatherNext is designed for enterprise, research, and meteorological agencies, not consumer-facing applications
  • User requires official weather warnings and alerts; WeatherNext is experimental and explicitly not an official weather report—users must rely on national meteorological authorities
  • User operates in regions outside current deployment areas; WeatherNext is actively deployed only with NHC, PAGASA, CWA, BMKG, VNMHA, with future expansion to Japan, Australia, and India
  • User needs historical weather data analysis; WeatherNext focuses on forward forecasting, not retrospective climate or weather pattern analysis
  • User lacks technical infrastructure for API integration; WeatherNext requires integration via Gemini Enterprise Agent Platform, BigQuery, or Earth Engine, not standalone consumer software
Commercials

Pricing

Pricing not specified