Jua is a weather intelligence company founded in 2022 by Andreas Brenner and Marvin Gabler in Zurich, Switzerland. Gabler previously led R&D at his family's meteorology firm and started Jua after growing frustrated with how slowly traditional weather science was adopting modern AI methods.
Rather than bolting machine learning onto legacy numerical weather prediction systems, Jua trains what it calls Large Physics Models: AI systems that simulate Earth's atmospheric physics directly, producing 20-day forecasts at 1x1km resolution with hourly updates. The company's own benchmarks show its EPT-2 model outperforming incumbents including ECMWF (the European weather agency that has led the field for four decades) and forecasting models from Microsoft, specifically on solar radiation and wind speed predictions that energy traders use to price procurement.
Jua is built for energy trading desks, utilities, and grid operators making procurement and pricing decisions around renewable output, not consumer weather apps. The company runs a free tier covering EPT-2 forecasts, ensemble data, and a Python SDK with meaningful API call allowances for development and testing, while production usage is billed per query at roughly $0.20 to $15 depending on the run, with full enterprise pricing requiring direct contact.
