AI-powered weather forecasting system by Swiss startup claimed superiority over Microsoft and Google's predictions
Revolutionary AI Weather Model Outperforms Competitors
In a groundbreaking development, Swiss startup Jua has unveiled its AI-powered weather forecasting model, EPT-2, which has been reported to outshine its competitors in accuracy, efficiency, and speed.
Jua, an active player in the field of AI-based weather forecasting, has developed EPT-2 as a native physics simulation model. Unlike traditional models that rely on complex equations and supercomputers, EPT-2 skips the equations and learns patterns from vast datasets, providing a more efficient and cost-effective approach to weather forecasting.
According to recent evaluations and public reports, EPT-2 has demonstrated superior performance compared to leading models such as Microsoft's Aurora, Google DeepMind's Graphcast, and the European Centre for Medium-Range Weather Forecasts (ECMWF)'s ENS and IFS HRES forecasts.
EPT-2 particularly excels in forecasts beyond 6 days for wind speed and maintains a temperature prediction advantage for nearly the full forecast range. The model uses 75% less computing power than Aurora, the second most efficient system tested, and processes forecasts 25% faster.
Moreover, EPT-2 has posted the lowest error scores of all models tested, capturing temperature peaks and short-term wind speed variations that Aurora missed, which are crucial for applications like energy management.
While the detailed academic validation is anticipated in forthcoming publications on arXiv and peer-reviewed venues, the initial evaluations and public reporting suggest that EPT-2 leads its competitors today.
Jua's CEO and co-founder, Marvin Gabler, is confident that EPT-2 can beat all competition, including DeepMind's Graphcast. The publication of the research comparing EPT-2 to other models is due to be published on the open-access archive arXiv next week.
AI-based weather forecasting has been making waves in recent years due to the demand for more accurate and cheaper ways to predict the Earth's climate. Jua has raised a total of $27mn in funding from backers including 468 Capital, Future Energy Ventures, and Promus Ventures.
Jua's first global AI weather model was released three years ago, and the launch of EPT-2 marks another significant milestone in the company's mission to revolutionise weather forecasting.
The revolutionary AI weather model, EPT-2, developed by Jua, demonstrates superiority in efficiency, accuracy, and speed over established models like Microsoft's Aurora, Google DeepMind's Graphcast, ECMWF's ENS and IFS HRES forecasts, and even outperforms DeepMind's Graphcast in certain aspects. EPT-2 significantly reduces computing power usage by 75% compared to Aurora, processing forecasts 25% faster. Furthermore, it exhibits the lowest error scores in temperature and wind speed forecasts, including short-term variations crucial for applications like energy management.
