> See the full index of documentation pages at https://docs.zettabyte.com/llms.txt.

# How billing works

> Prepaid wallet, what you are billed for, how charges accrue, and how to add funds.

Source: https://docs.zettabyte.com/docs/billing/how-billing-works

Last updated: 2026-07-21

Billing runs on a prepaid wallet, denominated in USD. You add funds by recharging the wallet, and usage is debited from the balance. The Billing page shows your current balance and an estimate of the amount consumed but not yet debited.

## What you are billed for

Usage is billed for GPU compute and for inference tokens. GPU capacity is purchased On-Demand and billed by the hour (see GPU instances), and inference token usage is metered separately (see Token usage). Itemized records for every charge are on Invoices and usage.

## How charges accrue

GPU billing is per resource allocation, hourly, and On-Demand. You are charged from the moment a resource is **allocated to you** until you release it, whether or not a task is actually running on it. Allocated but idle capacity still bills.

> [!WARNING]
> **Allocated resources bill until you release them**
> There is currently no pause for an instance. To stop charges you must **release the resource**, which returns the capacity to the pool. Anything not saved to persistent storage (a `PVC` or `S3`) is lost on release, so save your work first. See GPU instances and Storage.

Charges accrue as hourly billing events: while a resource is held, a billing event is recorded each hour. The confirmed billing basis is hourly, per allocation. Finer metering details for every resource type, and whether the exact per-hour amount is shown in the interface, are still being finalized.

## Wallet and balance

Your wallet holds a monetary balance in USD that your usage draws down. Top-ups are typically handled through your distributor or account owner rather than being fully self-serve, so plan ahead and request funds before your balance runs low.

> [!NOTE]
> **Two different balance numbers**
> Do not confuse your wallet balance with the balance shown in the resource queue. The wallet balance is *money*. The queue's balance appears to represent remaining GPU capacity, not funds. They are separate: a full wallet does not guarantee free capacity, and available capacity is not money.

## Whitelisted tenants

Whether you need to add funds at all depends on your tenant status. Whitelisted tenants are assigned resources and can use them directly, and may not need a bound payment method. Other tenants purchase resources before use. See Sign in and accounts.

## Payment and recharge

To purchase resources, bind a valid credit card to the account through a secure card-entry interface; you can set a default card for future transactions. Recharges add funds to the wallet and are listed under Recharge History, which shows the recharge time and amount.

## Pricing

zCLOUD does not publish resource prices in these docs. For current rates and longer commitments, talk to an expert.

## Related

- Invoices and usage: review transactions and export usage records.
- GPU instances: how GPU capacity is allocated, billed, and released.
- Token usage: track inference token consumption.