# How Cloud Compute is billed

> What an instance costs by what it uses, with memory while it runs, disk while it exists, CPU and data out, a worked example and the Paid plan.

Source: https://www.coritan.com/docs/cloud-compute/how-billing-works/

An instance you order on your own account is billed by what it uses, whichever size you choose. The size sets its vCPUs, memory and disk. It does not set a price. We measure use by the second, and take the cost from your credit each hour, or put it on your monthly invoice.

An instance you ordered before usage billing keeps the plan price it has until you [move it](/docs/billing/switch-to-usage-billing/). [How usage billing works](/docs/billing/usage-billing/) explains the parts every product shares, and [Usage rates](/docs/billing/usage-rates/) lists each rate.

## What you need to order {#needs}

- An instance needs the Paid plan. Pro, Business and Enterprise include it, and a Free account adds it ([The Paid plan](/docs/billing/usage-billing/#paid-plan)). Without it, the order answers `402` with the code `paid_plan_required`.
- A prepaid account needs the first deposit: $10.00 added in total, by default, before an instance runs. It stays in your account as credit. The order answers `403` until it is in. An account billed by invoice needs none ([What you need before you order](/docs/billing/usage-billing/#deposit)).

The order has nothing to pay up front, apart from any add-ons you order with it.

## What an instance pays for {#meters}

Memory
: Its memory size, for each second it runs. A paused instance still holds its memory, so it counts as running.

Disk
: Its disk, for as long as the instance exists, whether it runs or not.

CPU
: The processor time it uses, never the processor time it could use. If the host sends no reading for an hour, that hour has no CPU charge.

Data out
: What it sends to visitors, priced by the region of its location. Data it receives is free.

A Performance size pays 1.5 times the memory, CPU and disk rates.

## See the cost on the order page and the Billing tab {#where}

On the [order page](/docs/cloud-compute/create-an-instance/#choose-a-plan), each size shows its price an hour with every vCPU busy, a typical month and the most a month can cost, and the **Paid plan** card says whether your account has the plan. The figures are the `usage` block of the size, worked out from the rate card.

On the instance's **Billing** tab, the **Cost** card shows what the instance has cost **This month so far**, what **A typical month**, **Most a month**, **Idle, memory and disk** and **Stopped, disk only** cost for its size, whether the **Paid plan** is active, each rate, and a line for each meter with what it used, how much of it the Paid plan included and what it cost. The **Change plan…** button is not offered, because the instance has no plan price to change ([Change an instance's plan or cancel it](/docs/cloud-compute/billing/)).

## A stopped instance costs its disk {#stopped}

A stopped instance pays for its disk and nothing else. Its memory and CPU are not billed while it is off. It keeps its disk and its addresses, so start it again when you need it, or [cancel it](/docs/cloud-compute/billing/#cancel-the-instance) to stop the disk charge.

A stopped instance does not hold its memory on its host, so starting one is a new claim on the host. When the host has no room for the instance's memory, the start is refused with `There is no room to start this instance right now. Its disk is safe. Try again in a few minutes, or contact support if it stays like this.` The instance stays stopped and keeps its disk. We check this only for an instance billed by use. An instance that keeps a plan price has its place held ([Start, stop and restart](/docs/cloud-compute/power/#start)).

## Included each month {#included}

With the Paid plan, each month includes 25 GiB-hours of memory, 6.25 vCPU-hours of CPU, 200 GB-hours of disk and an amount of data out for each region, shared by all your instances, game servers and databases ([Usage rates](/docs/billing/usage-rates/#included)). Without the Paid plan, nothing is included and every unit is billed.

## A worked example {#example}

An `rx-2` instance has 2 vCPUs, 2 GB of memory and a 40 GB disk. It runs all month, a 730-hour month, and uses 10% of its vCPUs on average. It sends 500 GB out from a location in North America and Europe. The account has the Paid plan. These are the list rates and the included amounts of Oct 11, 2026.

| Meter | Used | Included | Billed | Cost |
| --- | --- | --- | --- | --- |
| Memory | 1,460 GiB-hours (2 GiB for 730 hours) | 25 | 1,435 at $0.0030 | $4.305 |
| Disk | 29,200 GB-hours (40 GB for 730 hours) | 200 | 29,000 at $0.00005 | $1.45 |
| CPU | 146 vCPU-hours (0.2 vCPU for 730 hours) | 6.25 | 139.75 at $0.015 | $2.096 |
| Data out | 500 GB | 1,000 GB | None | $0 |
| Total | | | | $7.85 |

The Paid plan adds its own price to the account once, however many instances the account runs.

If the same instance stays stopped all month, only its disk is billed: 29,000 GB-hours at $0.00005, or $1.45. If it is an `rx-2-performance` instance and runs the same way, memory costs $6.458, disk $2.175 and CPU $3.144, which is $11.78 in all.

A bill can differ from this arithmetic by a fraction of a cent, because each hour is rounded on its own.

## See what a size costs before you order {#estimate}

`GET /api/v1/products/` gives each size a `usage` block, worked out from its memory, vCPUs and disk at the current rates, before the included amounts:

```json
{
  "billing_model": "usage",
  "usage": {
    "hourly_idle": "0.0080",
    "hourly_typical": "0.0110",
    "hourly_max": "0.0380",
    "monthly_idle": "5.84",
    "monthly_typical": "8.03",
    "monthly_max": "27.74",
    "disk_monthly": "1.46"
  }
}
```

`hourly_idle` and `monthly_idle` are the cost with no CPU in use. The `typical` figures add CPU at 10% of the vCPUs, and the `max` figures have every vCPU busy all hour, which is the most the size can cost. `disk_monthly` is the disk part of a 730-hour month, which a stopped instance still pays. The `pricing` list in the same answer is the flat price brands and older services use. It is not what an order on coritan.com costs.

[`POST /api/v1/checkout/quote`](/docs/api/reference/client/checkout/#op-post-api-v1-checkout-quote) takes an order's body and answers `billing_mode` `usage`, no line for the instance, the same figures under `estimate`, and whether the order needs the Paid plan (`paid_plan`) or the first deposit (`deposit`).

## Add-ons {#add-ons}

An add-on that renews with the instance, such as the weekly and daily backups, costs a share of the instance's price, and an instance billed by use has none. The instance takes an add-on only when your account plan pays for it, as the Business plan does for automatic backups on up to three instances ([Automatic backups](/docs/cloud-compute/backups/#automatic-backups)). Floating IPs and the other products you order on their own are billed as they always were.

## Cap the cost {#cap}

There is no hourly cap on an instance, because it cannot use more than its vCPUs. Its `monthly_max` is the most a month can cost. An account [spend limit](/docs/managed-containers/how-billing-works/#spend-limits) counts every instance, and **Stop every instance** suspends them when the limit is reached. An account limit that is reached also refuses new orders of services billed by use.

## Related

- [How usage billing works](/docs/billing/usage-billing/)
- [Usage rates and included amounts](/docs/billing/usage-rates/)
- [Create an instance](/docs/cloud-compute/create-an-instance/)
- [Change an instance's plan or cancel it](/docs/cloud-compute/billing/)
- [How instance traffic is counted](/docs/cloud-compute/traffic/)