GenCast

Paid

Predicts weather and the risks of extreme conditions with state-of-the-art accuracy

4.6

New AI ToolsContact

Type

Saas

Founded

2010

Company

Google DeepMind

Visit Website

About GenCast

GenCast is a high-resolution (0.25°) AI ensemble model developed by Google DeepMind for probabilistic weather forecasting. It uses a diffusion model adapted to Earth's spherical geometry, trained on 40 years of historical weather data from ECMWF's ERA5 archive. GenCast generates an ensemble of 50 or more predictions, each representing a possible weather trajectory, providing a full probability distribution of future weather scenarios. In rigorous testing against the top operational system (ECMWF's ENS), GenCast achieved superior accuracy on 97.2% of 1320 forecast targets and 99.8% at lead times over 36 hours. It excels at predicting extreme events such as heat waves and strong winds, enabling better preparation. A single 15-day forecast can be produced in just 8 minutes on a Google Cloud TPU v5. Google DeepMind plans to open-source the model's code, weights, and forecasts.

Key Features

Probabilistic ensemble forecasts (50+ trajectories)

Up to 15 days ahead prediction

0.25° spatial resolution

Diffusion model adapted to spherical geometry of Earth

Trained on 40 years of historical weather data (ECMWF ERA5)

Outperforms ECMWF's ENS on 97.2% of forecast targets

Better prediction of extreme weather events (e.g., heat waves, strong winds)

Fast inference: 8 minutes on a single Google Cloud TPU v5 for a 15-day forecast

Planned open release of code, weights, and forecasts

Pros & Cons

Pros

State-of-the-art accuracy, outperforming the leading operational ensemble system
Probabilistic output captures forecast uncertainty effectively
Significantly faster than traditional numerical weather prediction (minutes vs. hours)
Better at predicting extreme weather events, enabling proactive measures
Open-source release of code, weights, and forecasts fosters community advancement

Cons

Relies on high-quality historical data (ECMWF ERA5) for training
Requires access to TPU hardware for fast inference (8 minutes on v5)
As a research model, not yet integrated into operational national weather services

Best For

Day-to-day weather forecastingExtreme weather risk assessment and preparationProbabilistic decision-making for energy, agriculture, and emergency managementClimate research and ensemble weather prediction

Alternatives to GenCast

PlugSugar

Automate conversations, answer questions with Web Search plugin, and customize ChatGPT experience using powerful AI plugins.

MedARC

Monitor patient progress, streamline processes and generate insights for healthcare providers to improve patient care.

SumUp

Experience Unmatched Sound Quality with AmbienceSoundPro!

Thinking Toolbox

Assess thinking skills, practice critical thinking, and access curated resources to improve cognitive abilities.

Respage

Automate lead acquisition, interact with potential leads, and capture lead information and preferences.

Travel Plan AI

Your personal AI guide for unforgettable journeys.

FAQ

What is GenCast?

GenCast is an AI ensemble model developed by Google DeepMind for probabilistic weather forecasting up to 15 days ahead, with state-of-the-art accuracy.

How does GenCast work?

GenCast is a diffusion model adapted to the spherical geometry of Earth. It takes the current state of weather as input and generates an ensemble of 50 or more possible future weather trajectories, representing the probability distribution of weather scenarios.

How accurate is GenCast?

In testing on 2019 data, GenCast outperformed the European Centre for Medium-Range Weather Forecasts' (ECMWF) ENS, the top operational ensemble system, on 97.2% of 1320 forecast targets, and on 99.8% at lead times greater than 36 hours.

Is GenCast open source?

Google DeepMind plans to release the model's code, weights, and forecasts to support the wider weather forecasting community.