Console
Compute
Dedicated GPU and CPU capacity for workloads that need more than the shared inference API.
Coming soonThis section previews work that is rolling out.
What to expect
Compute will let you reserve dedicated accelerators inside a project, with the same access, usage and audit model as the rest of the Console.Planned capabilities
- Reserve dedicated GPU and CPU instances per project.
- Bring your own container image and run long-lived or batch jobs.
- Per-project usage, cost and audit, consistent with inference and storage.
- Programmatic provisioning through the same API your Console uses.
Get early access
Talk to your UPGREAT contact to join the preview. As Compute ships, this page becomes the full guide and the endpoints appear in the API reference.