WeatherNext 3 is Google DeepMind's hourly weather model, and it reaches the public through several Google surfaces rather than a single product page. This guide maps those surfaces so you know which one answers your question. For the model's own detail, see architecture and resolution; for hands-on use, start with the step-by-step guide.
This is an independent experience, not a Google product. See plans

Google describes WeatherNext 3 as reaching Search, Gemini, Maps and its weather APIs, with research access through Earth Engine and cloud data products. Each path suits a different kind of question.
Ask about weather for a place and the answer can be informed by the model, in the same surface you already use.
Use a conversational request when you want context around the forecast rather than a fixed widget.
Developer-facing weather data for location products, where the model feeds the forecast layer.
Research and analysis paths for teams that want the fields themselves instead of an answer.
The right path depends on whether you want a quick answer, an interface, a pipeline or a dataset.
Search or Gemini when you need a forecast for one place and no further processing.
A browser workspace when you want to compare locations, variables and hours side by side with an account behind it.
An API when the forecast has to reach a dashboard, an alerting rule or an internal tool automatically.
BigQuery, Earth Engine or Cloud Storage when you are analysing history rather than asking about tomorrow.
National weather services remain the authority for safety-critical alerts, wherever the forecast comes from.
This site is a third-party service built on the model, with its own plans, credits and support.
What people ask about the Google connection specifically.
Related: API and data access and download options.
One account, hourly forecasts and the same interface on every device. Start the 7-day free trial, or read how the browser experience works first.