- Google has launched a new AI-powered forecasting model called WeatherNext 2.
- WeatherNext 2 makes forecasts faster, more accurately and at higher resolution, simulating hundreds of possible weather events in less than a minute.
- WeatherNext 2 now supports forecasts in Google Search, Gemini, Pixel and Maps.
Google revises your weather forecast with AI who thinks in probabilities. Instead of new radar towers or satellite launches, the new artificial intelligence-based forecasting model WeatherNext 2 developed by Google DeepMind and Google Research produces results up to eight times faster than traditional systems and can forecast hundreds of possible weather events from a single starting point at higher resolution and with greater accuracy than its predecessors.
WeatherNext 2 integrates with many of Google's most popular platforms, including Google Search, TwinsPixel Weather and Maps, with wider rollout coming soon via Google Maps Platform weather API.
What makes this update more than just a backend update is the scale of its ambition. WeatherNext 2 is designed to embrace uncertainty in unusual ways. Where older models might produce one most likely outcome, WeatherNext 2 can generate hundreds of potential futures, allowing forecasters and you to see the full range of possibilities.
This also means your forecast can't just say “Rain, 40% chance” but instead show multiple sequential results from your day's walk with a better understanding of what's actually likely to happen and when.
WeatherNext 2 uses what Google calls a functional generative network (FGN). The model does not rely solely on ready-made forecasts or entire weather systems; instead, it learns from individual, autonomous variables such as temperature, wind speed, and humidity. The model then determines how these variables interact to create “junctions”—complex real-world patterns such as storm fronts, heat waves, or regional changes in winds.
Google says this architecture allows WeatherNext 2 to outperform even its previous best-in-class model, providing more accurate forecasts for 99.9% of variables up to 15 days out.
It also makes forecasts much faster, completing a full forecast in less than a minute. By comparison, traditional physics-based forecasting can take hours on a supercomputer. This efficiency allows forecasts to be updated more frequently and in more detail.
Weathering of the future
After extensive testing, Google Gemini will begin to show forecasts based on WeatherNext 2 results, just like Google Maps. The average person could theoretically get a lot of benefits from the upgrade. Weather forecasting is one of those invisible systems that underlies a huge range of decisions. Make it more precise and you'll eliminate thousands of tiny sources of stress from people's daily lives.
There are also more serious consequences. For example, with more accurate weather forecasts, renewable energy providers can better estimate wind and solar output, and emergency services can respond with greater accuracy when forecasts reflect uncertainty rather than masking it.
This emphasis on uncertainty is key. The purpose of forecasting is not to be completely correct, but to prepare wisely for what may happen. By providing a range of physically realistic, interconnected scenarios, WeatherNext 2 pushes forecasting into something more strategic.
This may not solve the chaos of climate change and related natural disasters, but it could be a boon for those planning ways to better address these issues. AI-powered forecasting is starting to look like a necessary infrastructure.
Better data means better decisions. And this means much more to the weather than just helping us decide what coat we'll need in the morning.
Follow TechRadar on Google News. And add us as your preferred source to get our expert news, reviews and opinions in your feeds. Be sure to click the “Subscribe” button!
And of course you can also Follow TechRadar on TikTok for news, reviews, unboxing videos and get regular updates from us on whatsapp too much.
The best business laptops for any budget.
