Billing & Plans
How inference requests are funded, how you pay with QRIS or crypto, and the difference between Token Plans and PAYG.
Every inference request is paid for from prepaid credit you load onto your account. There are two funding sources, and zrouter spends them in a fixed order on every request.
How your requests are funded#
- Token Plan first. If you hold an active Token Plan with remaining tokens for the requested model, those tokens are consumed first.
- PAYG fallback. Any portion a Token Plan cannot cover is debited from your shared PAYG balance, priced at that model's PAYG rate. Rates are set per model — most are 90–96% below the official list price — and each one is listed on the pricing page.
- Insufficient funds. If neither source can cover the request, the gateway rejects it with
402 insufficient_balancebefore any inference runs.
You are charged after inference succeeds#
Funding is checked up front (admission), but the actual debit happens only after the upstream response settles successfully. A request that fails upstream is never billed. When a request draws from both a Token Plan and PAYG, the cost is split proportionally between them using integer arithmetic — there is no floating-point rounding.
How you pay#
| Method | Currency | Details |
|---|---|---|
| QRIS (tako.id) | IDR | You pay the product amount plus a 5% processing fee, shown clearly before checkout. Orders expire after 15 minutes, and confirmation is fail-closed — only a confirmed success credits your account. |
| QRIS — SumoPod | IDR | Zero processing fee — you pay exactly the order amount. Orders expire after 1 hour, and the minimum top-up is Rp5.000. |
| Crypto — USDC & USDT (Solana) | USD | Credited 1:1 to your PAYG balance for the amount actually received on-chain. Minimum top-up $2; an underpayment within 0.5% of the expected amount is accepted and credited for its received value (anything beyond that goes to manual review), and each account gets a dedicated deposit address. Funds are credited once the transfer is finalized — usually within tens of seconds. |
Token Plan vs PAYG#
| Token Plan | PAYG | |
|---|---|---|
| Scope | Per-model. A plan grants a fixed token allowance for one specific model. | Shared. One balance funds every model on the account. |
| Pricing | Prepaid tiers sold per model. Current tiers are listed on the pricing page. | Pay-as-you-go at a per-model rate per 1M tokens, shown next to the official list price on the pricing page. |
| Used when | You run one model heavily and want a predictable, model-specific allowance. | You use many models sporadically or want a single flexible balance. |
There is no subscription and no auto-renewal. A Token Plan simply adds model-scoped token allowance to your account, and PAYG credit is a shared spendable balance. Both are consumed automatically in the order described above.