The ECMWF AI Weather Quest is a new international competition designed to mobilise the AI and machine learning (ML) community to advance sub-seasonal weather forecasting, a scientifically complex but operationally critical area that covers the period from 2 to 6 weeks ahead. 

Organised by the European Centre for Medium-Range Weather Forecasts (ECMWF), a world-leading organisation in numerical weather prediction, the competition challenges participants to develop AI/ML models that generate probabilistic forecasts of temperature, precipitation, or sea-level pressure at the global scale. 

No meteorology background is required — the competition is open to all with AI/ML expertise, including participants from companies, startups, research groups, and independent developers. They are free to use any datasets and tools of their choice, and to innovate across model design, architecture, and approach. 

The primary goal is to benchmark the performance of AI/ML models for sub-seasonal forecasting, an area where traditional physics-based models face limitations. Forecasts will be submitted weekly and evaluated using a transparent evaluation system developed by ECMWF. 

The competition runs from March 2025 to at least September 2026, and includes an initial non-competitive training phase, followed by four 13-week competitive periods. Forecasts will be benchmarked, recognised in regular awards webinars, and showcased on a public portal. 

The ECMWF AI Weather Quest is supported by the European Commission as part of the Destination Earth initiative. 

Learn more and register: https://aiweatherquest.ecmwf.int.