This is the intended v1 offer for the hosted service once it launches. None of it, including the trial and checkout described below, is available yet.
plans
Standard is $20 per month. Pro is $50 per month. Both are subscriptions with a metered overage for hours past the included allowance.
Standard: 4 vCPU / 8 GB RAM, up to 8 parallel sandboxes sharing the workstation, 200 included active hours per month, $0.06 per extra active hour, default spend cap $40 per month, 40 GB workspace storage budget across sandboxes.
Pro: 8 vCPU / 16 GB RAM, up to 20 parallel sandboxes sharing the workstation, 300 included active hours per month, $0.12 per extra active hour, default spend cap $100 per month, 60 GB workspace storage budget across sandboxes.
An active hour is any hour in which the workstation is running. Stopped time does not accrue running-hour usage; subscription fees still apply. Overage counts completed hours only: a partial hour at the end of a period is never charged and never carried forward.
Parallel sandboxes share the workstation. The documented count is a ceiling, not a reservation of resources for each sandbox. Workstations become eligible for automatic idle pause after 24 hours with no attached client and no running job, subject to health and safety checks; stopping can take longer and other safety or billing stops may happen earlier. Running time, including idle time, counts toward usage. Storage is a shared managed budget, not a kernel-enforced quota: bytes your sandboxes hold on disk count toward it, including stopped sandboxes, and new work may pause at the budget.
These are the intended v1 offer. Your authenticated account and checkout display the terms offered to you; the accepted checkout terms govern your subscription. Nothing here reserves compute per sandbox or promises latency.
trial
The trial is 7 days with 10 active hours on Standard. A card is required. Unless you cancel before it ends, the subscription automatically converts to paid Standard at $20/month. If you use all 10 hours first, new compute is blocked until conversion; there is no trial overage or early charge. A trial is intended once per account. If you have used a trial or subscribed before, verify the terms displayed in Stripe Checkout before confirming: a trial request may instead offer immediate paid checkout. Choose Subscribe explicitly to pay without a trial.
If eligible, start a trial or subscribe from your authenticated client — pipod billing checkout --trial in the CLI for a trial request (omitting the flag is paid checkout), or the Start trial / Subscribe / Manage billing actions in the iOS and Android apps — which opens Stripe-hosted pages in the browser and returns here afterwards. This site never starts checkout. A card is required; trial eligibility is checked against recorded account history. Starting a workstation can take several minutes.
checkout and portal
Authenticated clients call POST /v1/billing/checkout-session for a Stripe checkout link and POST /v1/billing/portal-session for a Stripe billing-portal link (cancel and card update). Plan changes use the authenticated app's preview/confirm flow; a downgrade takes effect at period end. After either Stripe page, the browser returns to this site:
checkout return · checkout cancel return · portal return
Those pages confirm nothing by themselves. Account status updates after the server synchronizes with Stripe, via webhook or an authenticated refresh. The client re-reads the account with POST /v1/billing/refresh or its next status read. Anything after the ? in a return address is for Stripe's reference only and changes nothing on this site.
The spend cap limits new base-plus-metered commitments in the billing period, not taxes, prorations, adjustments or commitments already made before a cap was lowered. It is user-set, refuses new machine starts past the ceiling, refuses new pod work on an already-running workstation (launches, reuses, co-located runs, and wakes), and stops new overage reservations there. Every terminal billing state — an expired or exhausted trial, a failed payment past its grace period, a reached cap, or a missing subscription where one is required — blocks new starts and new work the same way. A reached cap additionally steers a running boat to sleep through the ordinary stop path. New work is refused at the limit. Existing work may pause after the remaining grace/TTL window. Stopping does not itself delete workspace files; retention and recovery limits still apply.
Canceling stops renewal, not current paid-period access or immediate data deletion. Billing does not block stopping a running workstation or reading files on a running eligible host; normal authorization and safety checks still apply. Fetching from a stopped workstation first needs a wake, and wakes require current eligibility, so a cold file fetch may have to wait until the account is in good standing. Self-hosted servers have no billing surface at all: under that edition the API serves no /v1/billing/* routes and carries no workstation billing block.