Compute setup guide

Check VM, cluster, GPU, and model runtime setup before choosing a compute workload.

Choose a compute workflow

  • Virtual Machines: create or inspect instances through the VM manager connected to this deployment.
  • Kubernetes: create or inspect clusters and workloads through the configured cluster manager.
  • GPU Resources: inspect the resources actually exposed by the selected endpoint.
  • Model Serving: invoke a registered model only after its executable artifact and inference runtime are ready.

Verify readiness first

Compute workspaces require account access and operator-configured execution infrastructure. Saving a resource request is not proof that an external instance or workload is running. Inspect status and runtime errors before using its address or output.

The /api/gpu/resources endpoint returns generated resources in development and an empty list in its current production branch. It is not a live cloud GPU inventory service.

Open the relevant guide

Need help with your setup? Visit the Help Center or contact Genorbex.