How billing works
This page explains the billing model. For the steps (top up, change plan, set a budget), see Manage billing and credits.
Two parts: a plan and a wallet
Billing is per organization and has two parts.
- Subscription. A monthly or yearly plan, priced per seat. It is the base fee for the platform and sets the markup applied to your usage.
- Credit wallet. Prepaid credits in US dollars. Everything you run is metered and paid from this wallet while it runs.
The split keeps the fixed part predictable and the variable part prepaid: you can never be surprised by a bill larger than what you put in the wallet.
Trial and signup credit
- Every new account starts a 14-day trial. No card is needed, and you can deploy during the trial.
- When the trial ends you have a short grace period of three days to choose a plan. After that you cannot deploy until you subscribe.
- Signing up with a verified email adds a one-time signup credit to your wallet. If you signed up with a wallet address, you receive it when you add and verify an email in Account settings.
What usage costs
Usage is billed at what the provider charges plus your plan's markup. Yearly plans carry a lower markup than monthly plans, and each line on your usage page shows both the raw cost and what you were charged.
- Compute is metered per hour of running time for each service. GPU and confidential (TEE) services are priced by the provider the same way.
- Prices differ by region and by GPU model.
acc regionsshows current availability and pricing before you deploy. - Closing a service stops its charges. A service that is failed or closed does not bill.
Paying
- Subscriptions are paid by card, or funded from your credit wallet if you prefer not to use a card.
- Credits are topped up by card in the web app, or with a stablecoin transfer (USDC, USDT, or DAI on Base, Ethereum, Arbitrum, Optimism, or Polygon) from the web app or the CLI.
When the wallet runs low
Two server-side safeguards protect the shared wallet:
- Before a deploy, the platform checks that your balance covers at least one hour of everything you would then be running (and at least one dollar). If not, the deploy is refused and you are asked to top up.
- While services run, if the balance falls below roughly one hour of total spend, services are suspended one at a time until the rest are covered. Suspended services are not deleted. They resume automatically after a topup.
Spend controls per service
Each service can carry its own limit, set when you create or deploy it and enforced by the platform rather than by your local machine:
| Mode | What it does | Flags |
|---|---|---|
| Pay as you go | No limit. The default. | --spend payg |
| Budget cap | Stops the service when its total or monthly spend reaches the cap. A monthly cap that the projected cost already exceeds is refused before launch. | --budget-total <usd>, --budget-monthly <usd> |
| Auto stop | Stops the service after a fixed runtime. The full projected cost is reserved from your balance up front. | --stop-hours <n>, --stop-days <n> |
When a service stops, the reason is recorded: budget exceeded, runtime expired, balance low, or manual stop.
Who can see billing
Owners and admins of an organization see the Billing page in the web app: usage, payments, invoices, plans, and the credits wallet. Members can deploy but do not see billing.
Next steps
Accounts, organizations, and projects
How an Alternate Clouds account is organized: organizations own billing and members, projects group services, and every service has deployments.
Docs for AI agents
How AI agents and LLM tools should read these docs and operate Alternate Clouds: llms.txt, Markdown endpoints, the Copy for AI button, the agent plugin, and the non-interactive CLI contract.