Release notes#

SimAI Pro 2026R1 allows you to train AI models to accurately predict new physical fields in minutes by leveraging previous 3D simulation data.

This application supports the following features:

AI Model Training

  • Selection and usage of local training data (VTP format)

  • Automatic random split of data into training and test subsets

  • Definition of model inputs such as scalar boundary conditions/parameters (JSON) or field inputs

  • Definition of output field variables

  • Specification of formulas for calculating global coefficients

  • Automated setup of the analysis domain

  • Definition of training duration

  • Optional downsampling of training data (beta)

  • Auto generated model evaluation report including quality metrics with HTML export

AI Model Prediction

  • Loading a geometry for a new design (STL, VTP format) and configuring boundary conditions/parameters

  • Running the AI model to predict results on surfaces

  • Postprocessing of predicted results with 3D viewer

  • Calculation of a confidence score to assess prediction quality

  • Exporting predicted results (VTP format)

  • 3D Visualization of the predicted results