Modeling layer
A consistent representation for geometry, materials, physical fields, boundary conditions, and study configuration.
Technology
Modern foundations for computational science.
Telion brings the modeling, numerical, automation, and assistance layers into one architecture. Each layer is designed to remain inspectable and extensible as the work becomes more specialized.
Architecture
Compute layer
Parallel numerical methods designed to take advantage of modern GPU hardware and scalable infrastructure.
Automation layer
A Python interface for custom models, parameter studies, optimization, data pipelines, and repeatable technical workflows.
Optimization layer
Differentiable runtime contracts for sensitivities, expression objectives, and certified parameter, shape, and topology workflows.
One model, from the desktop to Python.
The desktop interface and public Python API operate on the same versioned model representation. Geometry, materials, physics, studies, and results stay structurally aligned across both workflows.
CPU execution is always available. Compatible systems can opt into CUDA-backed assembly and linear algebra paths, with capability checks surfaced by the runtime.
Early access