CPU Instances
Manage CPU Instances
Control your CPU instance lifecycle — pause to reduce costs, resume to continue working, or terminate when done.
Instance statuses
CPU instances use the same status model as GPU instances:
| Status | Description | Billing |
|---|---|---|
PENDING | Instance is being provisioned. | No charge |
RUNNING | Instance is operational. SSH and JupyterLab are accessible. | Full compute rate (per-second) |
STOPPED | Instance is paused. Compute resources are released. Volume Disk and Network Volume data is preserved; Container Disk contents are not. | Reduced idle rate |
TERMINATING | Instance is shutting down. | Final settlement in progress |
TERMINATED | Instance is permanently shut down. All ephemeral data is deleted. | No charge |
FAILED | Provisioning failed or the instance crashed unexpectedly. | No charge |
Pausing
Click Pause on a running instance. The CPU resources are released and billing switches to a reduced idle rate. Any attached Network Volume data remains intact. Container Disk contents do not survive a pause — resuming provisions a fresh container, so installed packages and OS-level changes are lost.
SSH and JupyterLab become unavailable while paused. Use pause for overnight breaks or when you're not actively using the instance.
Resuming
Click Start on a stopped instance to resume it. CPU resources are re-allocated and your container restores with all previous data. Typically completes in under a minute.
Terminating
Click Terminate to permanently shut down the instance. All billing stops after a final settlement. The Container Disk is permanently deleted. Any attached Network Volume is not deleted and remains available for future instances.
⚠ This is irreversible. Download or back up your data before terminating if you don't have a Network Volume attached.
Billing lifecycle
Like GPU instances, CPU billing is per-second and settled on each status transition. Check the Deploy page in your dashboard for current compute and idle rates.