meterless Make-Good Policy
Version 1.3. Effective 6 October 2026. Part of the Terms of Service.
Paid subscriptions are not yet open. This policy applies to you from the day you subscribe.
This is what we do when the service falls short. It is automatic: our systems detect the failure and apply the credit. You do not open a ticket; our service records are the evidence.
1. The three commitments
| Commitment | What counts as a failure | What we do |
|---|---|---|
| Your studio starts | You ask for your studio and it is not ready within 30 minutes. Repeated requests during the same period of unavailability count as one failed start, timed from your first request. | We credit 1 hour to your plan for the failed start, plus 1 further hour for each full hour of delay beyond the first 30 minutes. |
| Your studio stays up | Your running studio is interrupted for more than 15 minutes through no fault of yours. | We credit 2 hours to your plan for the interruption, plus 1 further hour for each full hour it lasts beyond the first. |
| Your plan is provisioned | You subscribe, we take payment, and we cannot provide your studio within 24 hours. | We refund the payment in full, through Paddle to the payment method you used, within 14 days. You may keep the account and subscribe again later; the failed attempt costs you nothing. |
Credits are hours on your plan. Unused credit hours carry into the next month and expire at the end of it; carried hours do not count against that month's cap. New credits in any one month are capped at that month's plan hours: you can never be credited more in a month than the month you paid for. Credits are not cash and are not a per-hour price; meterless does not have one. If you are a consumer, nothing here limits a refund the law gives you.
2. What is not a failure
- A pause you asked for, or idle protection pausing an inactive studio.
- Your studio pausing because your month's hours are used.
- Time your workload spends queued or running slowly because of what you asked it to do (for example a large model loading, or a render that is simply long).
- Loss of connection on your side: your network, your browser, your device.
- Failures caused by software, nodes, models or settings you installed or changed.
- The availability of a third-party service you connect with your own account, or of a model run by a third party (Terms, section 8).
- Interruptions caused by your breaking the Terms or the Acceptable Use Policy, and a pause or suspension under the Terms or the Acceptable Use Policy, unless it proves unfounded.
- Planned maintenance we told you about at least 48 hours in advance, kept within the window we stated. We schedule maintenance for the quietest hours we can and we tell you by email and on our status page, whose address is on the help pages.
3. If we give you a substitute GPU
Your plan runs on a stated class of GPU. If we cannot place your studio on that class and run it instead on a substitute of a lower memory class, we tell you as soon as the substitution starts, on the studio screen and by email, naming the substitute and its memory plainly, and:
- for every hour you run on the substitute, we credit half an hour to your plan (degraded-but-working counts as half of not-working);
- if your work does not fit on the substitute (the studio reports an out-of-memory failure, or you tell us it will not fit with one click), your studio is paused, so no hours are used meanwhile, and every hour from then until we move you back counts as a full interruption under section 1, in place of the half-hour credits for those hours.
We never describe a substitute as an upgrade or a downgrade; we name the card class and the memory and let you decide.
4. If we move your studio
We may move your studio between machines of the same class to keep the service reliable. A move that keeps your studio available, or pauses it for less than 15 minutes, is not a failure. A move that takes longer is an interruption under section 1.
Our data-centre partners may sometimes reclaim a machine at short notice. That is our problem, not yours: we move you, and the interruption is credited under section 1 in the ordinary way.
5. If your data is affected
Your studio state is snapshotted when you pause and copied to encrypted storage. If, through our fault, a snapshot is lost and cannot be restored, we credit hours equal to one month of your plan's hours, outside the monthly cap in section 1, and we tell you what happened. This does not cover data you deleted yourself, or data on a studio that was never paused (and so never snapshotted). It also does not cover a snapshot after the storage period in the Terms has ended (30 days after your plan ends), because by then it has been deleted as the Terms say. Keep your own copy of anything you cannot afford to lose.
6. How credits are applied
Credits normally appear on your account within an hour of the failure ending, and in any case within three days, with a line saying what happened and when. Your account page shows every credit and its reason. If our systems miss a failure you experienced, tell us at support@meterless.com within 30 days, with the time and what you saw. We check it against our service records and credit any failure they show; if they show none, we tell you what they show, and a person will look again if you ask. If you are a consumer and the policy is unclear, it is read in your favour.
Credits stay with the account and do not transfer. If you cancel, you can use them until the end of your paid month. They lapse when the account closes.
7. When the failure is not ours
Sometimes a failure is caused by something outside our reasonable control: a data-centre power failure, a network provider outage, a natural disaster. We still credit under section 1, because you paid for the hours and did not get them. We do not use "circumstances beyond our control" as a reason to withhold these credits.
8. Relationship to your other rights
For consumers, this policy adds to the rights you have by law; it does not take any away, and any credit or refund we give counts towards any compensation the law gives you for the same failure. For business customers, this policy is the sole and exclusive remedy for failed starts, interruptions, substitute GPUs, provisioning failures and lost snapshots, as the Terms say.