Jua
Jua is a sophisticated tool that uses artificial intelligence (AI) for weather-dependent energy trading. The tool harnesses data, metrics, and deep neural netwo...
Last verified:
What is Jua?
Jua is a foundation model for reality that builds a high-resolution, real-time digital twin of Earth powered by AI physics models. It started with atmospheric prediction because learning coupled fluid dynamics, thermodynamics, and radiative transfer at planetary scale proves the model can learn anything. The platform delivers state-of-the-art earth systems forecasts (weather, climate, and more) optimized specifically for the energy industry, powering 100+ GW worldwide across five continents.
The Jua Platform brings together point forecasts from every leading weather model in one place, allowing users to compare signals side by side without switching tools. It includes Jua's proprietary EPT models (EPT 1.5, EPT 1.5 Early, EPT 2, EPT 2 Rapid Refresh updated 24×/day, EPT2e Ensemble 4×/day plus daily 60-day horizon), AI models from Microsoft Aurora, ECMWF AIFS, AIFS Ensemble, and DeepMind GraphCast, plus numerical models from ECMWF IFS, EC Ensemble, GFS, ICON, AROME, and GFS Ensemble. The platform offers market intelligence weighted for energy markets, interactive global weather maps, point forecast analysis, live model forecasting comparison, AI-enabled asset visualization, and Oracle for querying specific requests like trading nudges or renewable asset overviews.
Jua is designed for energy traders, utilities, renewable energy operators, analysts, developers, and data scientists who need extreme forecast accuracy for weather-dependent decisions. Energy companies that price power for five continents already run on Jua. The platform supports intra-day trading strategies, monitoring ensemble spreads, running demand models, energy procurement, renewables operations, and load forecasting. Developers can integrate forecasts into trading systems using the token-based API/SDK or Python SDK.
The EPT-2.0 flagship model delivers 20-day forecasts with hourly updates at 1x1km resolution with dynamic lead-time capabilities. The EPT-2e Ensemble provides 30-member probabilistic forecasting that outperforms ECMWF ensemble means. EPT-2 Rapid Refresh delivers hourly global updates powered by direct satellite observations, 12 times more frequent than traditional models. The system beats every incumbent forecast system on the metrics traders actually price, including RMSE, ACC, and CRPS.
Jua pricing
Pricing model: Free
Jua offers two subscription plans: Base Plan at €50/month includes essential AI weather models and numerical forecasting tools. Pro Plan at €500/month includes advanced AI models with market intelligence and real-time updates. The platform uses transparent token-based billing for API access where you only pay for data actually queried. Both plans include access to the web interface and API. Self sign-up is available with token authentication. Contact [email protected] or start a free demo for enterprise pricing and custom needs.
Jua pros
- 20-day forecasts at 1x1km hyperlocal resolution with hourly updates
- Beats ECMWF and all incumbent models on RMSE, ACC, and CRPS metrics
- 24×/day Rapid Refresh updates every hour for spot trading
- 30-member EPT-2e Ensemble for probabilistic forecasting
- All leading weather models in one platform - no tool switching
- Includes proprietary Jua EPT models plus Microsoft Aurora, GraphCast, ECMWF AIFS
- Token-based API/SDK with transparent usage-based billing
- Python SDK for easy integration with trading systems and analytics
- 60-day horizon ensemble forecast available daily
- Market intelligence weighted specifically for energy markets
- Oracle feature for custom queries like trading nudges and asset visualization
- Interactive global weather maps with point forecast analysis
- Live model comparison to see forecast signals side by side
- Capacity-weighted wind and solar predictions for renewables
- Population-weighted HDD/CDD for demand forecasting
- Used by utilities and trading desks across five continents
- Powering 100+ GW worldwide in production
- Built with ETH Zurich, Google Research, and ETH AI Center collaboration
- Self sign-up available with token authentication
Jua cons
- Primary focus on energy industry limits broader use cases
- Base Plan at €50/month may be expensive for individual users
- Pro Plan at €500/month targets enterprise only
- Historical hindcast data available only on request
- No publicly documented free tier for testing
- Complex platform may have steep learning curve for non-traders
- Limited public documentation compared to mainstream weather APIs
- Primarily serves professional traders rather than consumers
- Newer company with less track record than established weather providers
Frequently asked questions about Jua
What is Jua's EPT-2 model and how is it different from other weather models?
EPT-2 is Jua's flagship proprietary AI weather model that delivers 20-day forecasts at 1x1km resolution with hourly updates. It is a Large Physics Model trained on petabyte-scale data as an end-to-end physically consistent system. Unlike traditional numerical weather prediction models that run on massive supercomputers, EPT-2 is 100% AI-powered and significantly more accurate, beating ECMWF IFS and all incumbents on RMSE, ACC, and CRPS metrics that energy traders actually price.
How does the Jua Platform work for energy traders?
The Jua Platform brings together point forecasts from every leading weather model in one interface so traders can compare signals side by side without switching tools. It provides market intelligence weighted for energy markets, interactive global weather maps, live model forecasting comparison, and AI-enabled asset visualization. The platform updates continuously and includes Oracle for custom queries like setting up trading nudges or generating visualized overviews of renewable energy assets.
What weather models are available through Jua?
Jua provides access to three categories of models: Jua proprietary models (EPT 1.5, EPT 1.5 Early, EPT 2, EPT 2 Rapid Refresh updated 24×/day, EPT2e Ensemble 4×/day plus daily 60-day horizon), AI models (Microsoft Aurora, ECMWF AIFS, AIFS Ensemble, DeepMind GraphCast), and numerical models (ECMWF IFS, EC Ensemble, GFS, ICON, AROME, GFS Ensemble). All models are accessible through the platform, API, or Python SDK.
What is EPT-2 Rapid Refresh and when is it useful?
EPT-2 Rapid Refresh is Jua's fastest global forecast, updated 24 times per day every single hour. It is powered by direct satellite observations and delivers real-time accuracy far beyond traditional models that refresh only four times daily. For spot energy traders, Rapid Refresh means staying ahead of sudden weather shifts, optimizing intraday positions, and making confident calls in volatile markets. It provides up to 2.5 hours forecast advantage before the market.
What is the EPT-2e Ensemble model?
EPT-2e is Jua's state-of-the-art 30-member ensemble model designed for fully probabilistic forecasting. It clearly outperforms long-standing benchmarks like HRES and ECMWF ensemble means, redefining accuracy and reliability in medium- to long-range weather predictions. For energy traders, this means sharper demand forecasts, better risk hedging, and stronger confidence in long-horizon positions. It is updated daily and available via platform, API, or SDK.
How do I integrate Jua forecasts into my trading system?
You can integrate Jua forecasts using the Python SDK with pip install jua, which provides direct access to all Jua models and partner models through a simple token-based API. The SDK gives programmatic access to point forecasts, capacity-weighted wind and solar, and population-weighted HDD/CDD. You can also use the OpenAPI endpoints for direct integration with any coding language. Transparent token-based billing ensures you only pay for data you query.
What is Oracle on the Jua Platform?
Oracle allows users to query very specific requests, unlocking new dimensions for individual modification and a broad range of different use cases. From setting up trading nudges to generating a visualized overview of your renewable energy assets, with Oracle you can do it all. It is now available on the Jua Platform and enables custom AI-enabled queries beyond standard forecast data.
Who is Jua designed for?
Jua is designed for energy traders, utilities, renewable energy operators, analysts, developers, and data scientists who need extreme forecast accuracy for weather-dependent decisions. The companies that price power for five continents already run on Jua. It is perfect for managing intra-day trading strategies, monitoring ensemble spreads, running demand models, energy procurement, renewables operations, and load forecasting. Developers use it for integrating weather data into trading systems, analytics workflows, or custom applications.
What resolution and update frequency does Jua provide?
Jua's EPT-2.0 model delivers forecasts at 1x1km hyperlocal spatial resolution with hourly updates for 20-day horizons. EPT-2 Rapid Refresh updates 24 times per day (every hour) for the fastest global forecast. EPT-2e Ensemble updates 4 times per day plus a daily 60-day horizon forecast. This provides 12 times more frequent updates than traditional weather models for critical high-frequency weather data.
How accurate is Jua compared to other weather models?
Jua beats every incumbent forecast system on the metrics traders actually price, including RMSE (root mean square error), ACC (anomaly correlation coefficient), and CRPS (continuous ranked probability score). The EPT-2 model is significantly more accurate than leading numerical models, providing accurate results for up to 48 hours into the future where traditional models only manage 12 hours. It outperforms Big Tech models from Microsoft, Google, and Nvidia, as well as ECMWF benchmarks.