Release Notes
This document outlines the features and improvements included in each release of the Jua platform.
Latest Release 2.9.2025
Jua EPT-2.0 Rapid Refresh (Research Preview)
Hourly Updated Global Model - Released Jua EPT-2.0 Rapid Refresh, a research preview global weather model updated 24 times a day (every hour) with direct satellite observations.
Industry First - Unlike existing numerical and AI weather models (which update only 4 times a day using numerically assimilated conditions), Rapid Refresh provides real time updates and high accuracy.
60-Day Ensemble Forecast EPT2e
New Long-Range Forecasting - Introduced a 60-day probabilistic ensemble forecast that outperforms the ECMWF Ensemble Mean (EC Ens).
Daily Updates - Forecasts are updated every day by default, with the option for more frequent updates upon request.
Paradigm Shift - Fully probabilistic and designed for long-horizon decision-making, setting a new standard in ensemble-based forecasting.
Jua Maps: Next-Generation Visualization
Browser-Based Visualization Platform - Launched Jua Maps in the Jua platform, providing a high-performance tool for exploring and animating weather data.
Interactive Weather Layers - Supports visualization of wind, precipitation, temperature, geopotential, isobars, pressure, and cloud coverage.
Model Coverage - Available for Jua EPT-2, EPT-1.5, EPT-1.5 Early, ECMWF AIFS, and Microsoft Aurora.
Fast & Performant - Built for responsive, smooth animation and exploration directly in the browser.
Jua Market Aggregates 2.0
Expanded Energy Forecasting Tools - Released an updated version of Jua Market Aggregates with capacity-weighted wind and solar forecasts.
Cross-Model Comparisons - Covers all Jua models and major third-party models including EC, EC Ens, GFS, ICON, AIFS, and more.
Advanced Analytics - New tools include graphs, deltas, disagreements, and heatmaps, enabling deeper analysis of market impacts from weather variability.
Release 29.6.2025
Platform Accessibility & Enhanced User Experience
This release focuses on making the Jua platform more accessible to new users while expanding model availability and improving forecast visualization capabilities.
Self Sign-Up & Simplified Access
Self Sign-Up Enabled - Users can now create accounts independently without requiring manual approval, streamlining the onboarding process for new customers and trial users
Usage-Based Pricing with Tokens - Introduced a usage-based pricing option based on tokens (like LLM tokens), providing a more flexible and transparent cost structure for API consumption
Token Usage Calculator - Added a built-in calculator that helps users estimate API consumption costs and track usage patterns before committing to API calls
Improved User Onboarding
Guided Onboarding Flows - New users are now guided through interactive tutorials that demonstrate key platform features, model selection, and basic forecasting workflows
Progressive Feature Discovery - The onboarding process introduces advanced features gradually, helping users understand the platform's capabilities without overwhelming them initially
EPT-2e Ensemble Model Deployment
EPT-2e Ensemble Forecasts - Deployed our probabilistic ensemble version of EPT-2, providing uncertainty quantification and confidence intervals for all forecasts
Ensemble Statistics - Full statistical analysis available across the dashboard, Python SDK, and API endpoints, including percentiles, standard deviation, and confidence bands
Updated Documentation - Comprehensive documentation updates explaining ensemble interpretation and best practices for probabilistic forecasting in energy trading applications
Enhanced Forecast Visualization & Control
Historical Initial Conditions - Users can now select any forecast initialization time from the past 36 hours, enabling comparison of how forecasts evolved and analysis of forecast consistency over multiple model runs
Live vs. Cached Forecasts - Toggle between the latest available forecast and any cached forecast from recent initialization times, perfect for tracking forecast changes and model performance
Dynamic Confidence Bands - Interactive confidence interval visualization that adjusts based on selected ensemble percentiles, providing visual uncertainty quantification for any forecast period
Expanded Model Portfolio
EPT-2 Early Model - Deployed the early-release version of EPT-2 that provides forecasts 2.5 hours ahead of the standard EPT-2 timeline, crucial for time-sensitive trading decisions
Microsoft Aurora Integration - Added Microsoft's state-of-the-art Aurora AI weather model to the platform, providing additional model diversity for ensemble-based decision making
ECMWF AIFS Platform Access - ECMWF's Artificial Intelligence Forecasting System is now fully integrated into the platform interface, complementing existing API access
User Experience Improvements
Global Timezone Settings - Users can now set their preferred timezone in account settings, with all timestamps and forecast displays automatically adjusted to their local time zone, eliminating confusion around UTC conversions
Release 25.4.2025
New Flagship EPT-2 Model & Power Trading Analytics
This release introduces our new state-of-the-art EPT-2 model and the first version of Jua power trading analytics.
EPT-2: Our New Flagship Model
EPT-2 - State-of-the-art weather model outperforming leading public AI weather models including Microsoft Aurora, DeepMind's GraphCast, ECMWF's AIFS, and our previous EPT-1.5 model
Hindcasts for EPT-2 - Historical forecast data available for performance analysis and evaluation
Hindcasts for ECMWF AIFS - Historical forecast data now available for ECMWF's AI model
Power Trading Analytics (Beta)
The first version of Jua power trading analytics is now available for selected beta users, including:
Power Forecasts - Energy production forecasting based on weather data
Capacity Analysis - Renewable energy capacity insights and predictions
Release 12.3.2025
New weather models & API enhancements
The latest release significantly expands the available weather models and enhances the API functionality.
Available Weather Models
The following models are now available through both the platform interface and API:
ECMWF Operational - ECMWF IFS HRES model, often referred as ops
ECMWF Ensemble Mean - Average of multiple ECMWF IFS Ensemble runs
ECMWF AIFS - EC's new AI model
GFS - Global Forecast System from NOAA
GFS Ensemble Mean - Average of multiple GFS ensemble runs
AROME France HD - Model from France Meteo
ICON-EU - Model from DWD - Deutscher Wetterdienst
Enhanced API Endpoints
The second release introduces the init_time parameter, allowing users to query forecasts from specific model initialization times. This enables:
GET /v1/forecasting/model_name/forecasts/init_time Retrieves forecast metadata for a specific model initialization time.
GET /v1/forecasting/ecmwf-operational/forecasts/20240310T000000ZHistorical forecast comparison
Forecast accuracy analysis over time
Access to specific model runs for research or operational purposes
Format: YYYYMMDDTHHMMSSZ (ISO 8601 format with Z indicating UTC time)
Using the New Models
To access the newly added weather models, simply specify the model name in your API requests:
# ECMWF Operational
GET /v1/forecasting/ecmwf-operational/forecasts/latest/47.37,8.54
# ECMWF Ensemble Mean
GET /v1/forecasting/ecmwf-ensemble-mean/forecasts/latest/47.37,8.54
# GFS
GET /v1/forecasting/gfs/forecasts/latest/47.37,8.54
# ECMWF AIFS
GET /v1/forecasting/ecmwf-aifs/forecasts/latest/47.37,8.54
# GFS Ensemble Mean
GET /v1/forecasting/gfs-ensemble-mean/forecasts/latest/47.37,8.54
# AROME France HD (regional model)
GET /v1/forecasting/arome-france-hd/forecasts/latest/48.85,2.35
# ICON-EU (European regional model)
GET /v1/forecasting/icon-eu/forecasts/latest/52.52,13.40Initial Release of Jua Platform
First deployment of new home of Jua data, analytics & tools
The initial release of the Jua platform included:
Launch of the main application at app.jua.ai
Introduction of core platform features:
Weather forecast visualization interface
Comparison of forecasts from multiple models
Location-based weather data retrieval
Customizable forecast parameters
Key Features
/benchmarking - Performance comparison against ERA5 reanalysis data
/documentation - Comprehensive API and usage documentation
/legal - Terms of service and legal information
/forecast - Interactive REST API testing tool
Forecast Interface
The forecast interface allows users to:
Select locations through search or from preset city options
Configure forecast horizons (up to 480h)
Choose from multiple weather models
View customizable parameters (temperature, precipitation, etc.)
Compare model performance live through interactive charts
Last updated