What are AI credit overages? Billed later or refused
AI credit overages are usage past a plan's included credits, billed later or blocked until you buy more. Sume refuses the job with a 402 until funds are added.

AI credit overages are usage beyond the credits your plan includes. A tool that allows overages keeps running and bills the extra afterwards; a tool without them stops the work until you buy more credits. On Sume the balance is prepaid: when it can't cover a generation's estimated cost, the request fails with 402 insufficient_credits before anything runs.
The Sume side comes from its Terms of Service (last updated August 10, 2026), the pricing FAQ, the Generation admission and Errors docs, and current billing code, read on 2026-09-29.
How do AI credit overages work?
There are two models. Read the plan page for which one a tool uses:
- Postpaid overage. Work continues past the included credits, and the extra appears on your next invoice at the overage rate. Nothing stops mid-project, but the bill can grow while nobody is watching.
- Prepaid only. Work past the balance is refused until someone adds funds. The bill can't exceed what you paid in, but a job can fail at a bad moment.
- An "enable overages" switch, where a tool has one, moves you from the second model to the first. Before you turn it on, check the overage rate, whether there is a cap, and who in your team can change it.
What happens on Sume when the balance runs out?
Sume works the prepaid way. The terms say each plan includes a prepaid generation allotment and that "work may be declined when a balance is exhausted". Generation admission checks the spendable USD balance at submit: if it can't cover the reservation, the submit "fails with 402 insufficient_credits before provider work starts". Fixing that error step by step is covered in Insufficient credits error 402.
| Situation | What Sume does |
|---|---|
| Balance can't cover a generation | Submit fails with 402 insufficient_credits before provider work starts |
| Adding funds | One-time purchases from $10 to $1000 into a team's shared wallet |
| Who can add funds | Organization admins only, in current code |
| Does buying more raise parallel jobs? | No: generation concurrency is plan-only; top-ups don't raise it |
| Cap on one Format run | generation_spend_cap_usd, up to the platform maximum of $500 |
| Custom billing | Enterprise: custom terms or invoice billing |
Can I enable overages on Sume?
Sume's self-serve terms describe prepaid balances, not billed overages. To keep work running past the included usage, add funds ahead of time:
- Teams fund a shared wallet with on-demand purchases from $10 to $1000, per the pricing FAQ. In current code, only organization admins can add funds, and the personal top-up route answers
stripe_topup_team_only. - Or upgrade: the
stripe_topup_team_onlymessage in current code suggests upgrading your plan "for more included usage". - Larger organizations that need custom terms or invoice billing are pointed to Enterprise. A signed Order Form can set billing schedules, commitments and credit treatment.
How do I stop usage from running past my budget?
A prepaid balance is itself a ceiling, but a large balance can still be spent sooner than you meant. Two controls narrow it:
- Spend approval for agents. The terms say Sume asks agents to confirm spend before paid work, and that you can approve every generation, approve automatically up to a dollar threshold you set, or let runs proceed without approval. The setting also authorizes spend by scheduled runs.
- A per-run cap. A Format run takes
generation_spend_cap_usd, "up to the platform maximum of $500" (Call a Format).
What else should I check on any plan with credits?
Ask whether bought credits expire and whether unused included credits roll over; on Sume, the answers are in How do AI credits work?. Ask whether buying more raises how many jobs run at once; on Sume it does not, since concurrency is set by the plan. How the plans, the one wallet and published model rates fit together is in How Sume pricing works.
Sources
Related posts
More in Pricing
- AI dubbing rate per minute: a step-by-step cost breakdown
An AI dub's rate per minute is the sum of its steps: transcribe, translate, speak, render. Sume's per-step prices and a worked 1- and 10-minute dub.
- AI image generator for commercial use: what the terms allow
Commercial use of AI images is set by each tool's terms and plan, not the model. On Sume, paid plans include it and Sume claims no ownership of outputs.
- AI music generator cost: what one track costs to make
An AI music generator API costs a flat price per generated track on Sume, not per minute. The rate, what changes it, and budgets for takes.
- AI video generator for enterprise: terms, billing, limits
Buying an AI video generator for an enterprise means a signed order form, invoice or PO billing and higher limits. What Sume's Enterprise tier lists.
Written by Sume