CPU Instances
Deploy a CPU Instance
Launch a general-purpose CPU cloud instance — ideal for data processing, scripting, and lightweight workloads.
Step 1: Open the CPU Deploy tab
Navigate to Dashboard → CPU Compute and click the Deploy tab. This loads the live CPU catalog — a listing of available CPU flavors with their vCPU counts, RAM allocations, and per-hour rates.
Step 2: Select a CPU flavor
CPU flavors are grouped by their RAM-to-vCPU ratio. Each flavor offers multiple tiers — preconfigured combinations of vCPU count and RAM. For example, a flavor with 4 GB/vCPU might offer tiers of 2 vCPUs / 8 GB, 4 vCPUs / 16 GB, up to 24 vCPUs / 96 GB.
Click on a flavor card, then select the tier that matches your workload requirements. The page scrolls to the deploy configuration panel.
Step 3: Choose a template
Click Change Template to browse container images optimized for CPU workloads. The default template includes Python, common data science libraries, and a lightweight base OS.
Templates are categorized (Official, Community, etc.) and show the included software stack. Select the one closest to your needs.
Step 4: Configure storage
CPU instances use a Container Disk for ephemeral storage. The default size is 20 GB, adjustable via the slider during deployment. You can also optionally attach a Network Volume for persistent storage.
For details on storage behavior and persistence, see Network Volumes.
Step 5: Access options
Both SSH and JupyterLab are enabled by default — the same setup as GPU instances. SSH requires a public key saved in Account Settings → Security.
Step 6: Set a pod name and deploy
Give your instance a name (auto-generated by default), review the configuration, and click Deploy. Your balance must cover at least one hour of the selected tier.
The instance appears in the My Instances tab with PENDING status, transitioning to RUNNING once provisioning completes.