Platform
Local runners versus cloud runners: which one you actually want
The agent runs CAD tools on a machine. Whether that machine is yours or ours changes cost, throughput and where your geometry lives.
Aikyora · Jun 10, 2026 · 5 min read
The agent does not simulate CAD tools. It runs them: Blender, Rhino, FreeCAD, CadQuery, PyMeshLab and ParaView, headlessly, on a machine called a runner. The interesting question is which machine.
A local runner keeps geometry on your hardware
The work happens on a workstation you control, and the geometry never leaves it. Runtime costs you nothing beyond the machine you already own, which is why the entry plans include local runners only.
A cloud runner buys throughput
Cloud runners let several jobs proceed at once and remove the dependency on one desk being switched on. Runtime there is metered by the second, so it shows up in credits.
On-premise is a third answer
For work that cannot leave a network at all, Enterprise deployments run the agent against on-premise runners. Same scripts, same outputs, different boundary.