Google’s latest AI weather model gives you no excuse to forget your umbrella

Source: TechCrunch AI By Tim Fernholz Updated
Image: TechCrunch AI

Scientists at Google DeepMind and Google Research released a new artificial intelligence model for weather forecasting today that sees our changing atmosphere more clearly and predicts its behavior more often. WeatherNext 3 is the latest wave of a sea change in meteorology brought out by deep learning techniques, and Google says it will start feeding…

Opening of the original on TechCrunch AI

Summary

Google's new AI weather model, WeatherNext 3, achieves the highest accuracy on the Operational WeatherBench benchmark, surpassing traditional forecasts and other AI models. Developed by Google DeepMind and Google Research, it offers improved resolution down to 5 km, better rain prediction, and hourly forecasts. This model will integrate into Google Search, Maps, and Gemini, and will be available on Google Cloud platforms, marking a significant step in AI-driven meteorology.

Why it matters

This release marks a significant advancement in AI weather forecasting, directly competing with and surpassing traditional supercomputer models. WeatherNext 3's improved accuracy and resolution will enhance consumer-facing products like Google Search and Maps, and offer researchers more powerful tools via Google Cloud. While Google claims it's the first to incorporate raw observational data for high-resolution global forecasts, WindBorne has a similar claim. The key differentiator for Google appears to be global resolution. This development could democratize access to accurate weather data, especially for regions lacking expensive infrastructure, and signals a continued shift from physics-based to data-driven weather prediction.

Read this on TechCrunch AI
Opens in a new tab. Subvolts summarizes and links; the full piece belongs to TechCrunch AI.
Where the other five stand

Related: OpenAI: Introducing GPT-6 Astra: the most intelligent and aligned model in the world. · Anthropic: India Country Brief: The Anthropic Economic Index · Microsoft: Upcoming deprecation of selected GitHub Copilot models · Meta: An Organizational Second Brain: Building an AI That Learns From Experts · xAI: Sam Altman "AGI by December"

Hype check
4/5Big

Rated high: a launch, deal or policy change that changes what you can do or what it costs.

Who's talking about it
Prior coverage our earlier items on the same thing
Published
Updated
Sep 3, 2026
Source
TechCrunch AI (techcrunch.com)
Author
Tim Fernholz
Company
Google · Web · News
People
Samier Merchant, Ferran Alet, Daniel Rothenberg
Products
WeatherNext 3, Gemini, WeatherNext 2, WeatherMesh 6
Summary by
Subvolts, using an AI model (how we work). Spotted a mistake? Tell us.

Questions people ask

How accurate is Google's new AI weather model?
WeatherNext 3 is the most accurate among leading contenders tested on Operational WeatherBench. It beats other deep-learning models and traditional forecasts from government agencies.
What improvements does WeatherNext 3 offer over its predecessor?
It predicts down to 5 km resolution, improves rain evaluations by 60%, and provides hourly forecasts instead of every six hours. It also incorporates raw observational data directly.
Where will users see WeatherNext 3's forecasts?
The model will feed into Google Search, Google Maps, and Gemini. It will also be available to users and researchers on Google's cloud platforms.

More from TechCrunch AI

This is the first item Subvolts has from TechCrunch AI. More arrive with each daily run.

Page generated Sep 3, 2026. Summaries are Subvolts' own; the story belongs to TechCrunch AI.