Swiss startup Jua has introduced EPT-2, an AI-powered weather forecaster boasting higher accuracy and speed compared to major models from Microsoft and Google DeepMind. EPT-2 reportedly delivers superior forecasts on critical parameters such as wind speed and air temperature while being 25% faster than its closest competitor, Aurora. Jua claims its model operates with 75% less computing power than Aurora, positioning it as a more efficient option. The results are slated for publication on arXiv, and CEO Marvin Gabler expresses confidence in EPT-2's capabilities against prevailing models in the industry.
Jua's EPT-2 model not only outran leading competitors in 10-metre wind speed and 2-metre air temperature accuracy but also executed forecasts 25% faster while consuming 75% less computing power than Microsoft's Aurora.
Recent comparisons showed EPT-2 outperforming Microsoft's Aurora and the European Centre for Medium-Range Weather Forecasts, which are typically considered the industry benchmarks for weather forecasting accuracy.
CEO Marvin Gabler states that Jua's model surpasses competitors by being faster and more efficient, challenging established systems that are slower or reliant on outdated infrastructure.
The launch of Jua's model represents a significant leap for AI in weather prediction, fulfilling a heightened demand for accurate and economical forecasting methods.
Collection
[
|
...
]