Skip to main content
Every organization is on exactly one plan. The plan sets three things: how much usage is included, how large a single service may be, and which capabilities are available. Compute is billed on top of the plan at the same rates regardless of tier. A plan is a set of ceilings and capabilities, not a different price per vCPU.

What each plan includes

Rates

Compute is metered by the second against the allocation a service is configured with, so a service that sleeps costs nothing while it is asleep. Disk, backups and object storage are metered on what is provisioned or held, whether or not the service is running.

How limits are enforced

Limits behave differently depending on whether you are choosing something or shipping something. Choosing is refused. Setting a service above your plan’s CPU or memory ceiling in the console, or adding a seat beyond the seat limit, returns an error and nothing changes. Shipping is clamped. If a slipway.yaml asks for more CPU or memory than the plan allows, the deploy still runs and the allocation is trimmed to the ceiling. slipway records a resources_clamped event on the deployment so the trim is visible. A plan downgrade therefore never turns a push-triggered deploy red. A downgrade never destroys anything. Losing a capability stops you creating new things that need it. Buckets, previews, disks and gated services that already exist keep working.

Capability notes

Always-on services. On Free, a web service with a public port sleeps after five idle minutes and wakes on the next request. Services that cannot receive HTTP, such as workers and databases, do not sleep and run continuously. Shell, scaling and disks. These are workspace capabilities rather than properties of a service size. On a plan without them, the console shows the tab with a lock and the API refuses independently. Object storage and backups. Both are gated at creation only. Database high availability and connection pooling. Both are gated at the enable toggle and at deploy. High availability requires Pro or higher; connection pooling is included on every paid plan at no extra charge. A downgrade stops enabling either on new databases and never touches a cluster or pooler that is already running. Bring your own cluster. Registering a new cluster requires Pro or higher. Clusters already registered keep running if the plan changes.

Running out of balance

When an organization’s balance reaches zero, projects pause. Nothing is deleted, disks and their contents are kept, and adding balance resumes them. See Billing.

FAQ

Can I change plan at any time? Yes. Changing plan takes effect immediately, and usage already metered is unaffected. Does the Free grant renew? No. It is a one-time $5 credit per organization. Paid plans include a monthly usage credit instead, which does renew. Is the Pro credit per seat? No. The 20monthlycreditisperworkspace.The20 monthly credit is per workspace. The 20 price is per seat. Do previews cost extra? No. A preview is billed as the services inside it, at the same rates, for as long as they run. Previews themselves carry no separate charge.