Pricing

Two prices, both transparent:

A 9% markup on inference. When you make a call, we pay the underlying vendor, add 9%, and charge you the total.
A service fee on top-ups. When you add credit to your account, we charge a small service fee on the transaction.

There is no subscription. There is no monthly minimum. You pay only when you make calls.

The 9% inference markup

Each request is priced as:

plaintext

customer_charge = round(vendor_cost × 1.09)

We use the vendor's published rate card for the model you called. The vendor's cost is computed from the actual token usage they return. We add 9% on top and debit your balance for the total. Each request appears in your dashboard with both numbers — what the vendor charged us, and what we charged you.

Why 9%. It funds the infrastructure (routing, the prompt-extraction guard, the dashboard, the logs, support) at a rate that's roughly half of the spread most major aggregators charge. We considered going lower; the math at our scale didn't allow it without compromising on the platform's reliability work.

Why this is still a savings. COP — our proprietary optimization layer — reduces your output token count enough that, even with the 9% markup, the total bill on most workloads is below what you'd pay going direct. The marketing claim is "Save up to 15%":

Coding workloads see roughly 24% fewer output tokens. After the 9% markup, the net savings clear 15%.
Mixed conversational workloads see 5–15%.
Some workloads (raw text generation with no structural overhead) see less, occasionally near zero.

Your dashboard shows the actual per-call delta for your traffic — not an estimate, not an average. The number on your screen is what your queries actually saved.

The service fee on top-ups

Your account holds a credit balance. You add funds via a top-up. On each top-up, we charge a service fee in addition to the credit amount. The fee structure is hybrid:

Below $60.18: 3.5% of the credit amount + $0.30 flat.
$60.18 and above: 3.5% of the credit amount, flat (no $0.30 add-on).

The threshold is set at the exact break-even point of the 3.5% rate against payment-processor cost, so the structure stays margin-positive at every dollar from $5 up. There is no "loss zone" where we eat the fee, and there is no "trap zone" where you pay more proportionally than you should.

Top-up examples

| You top up | Service fee | You pay at checkout | Credit you receive | |---|---|---|---| | $5 | $0.48 | $5.48 | $5.00 | | $10 | $0.65 | $10.65 | $10.00 | | $25 | $1.18 | $26.18 | $25.00 | | $100 | $3.50 | $103.50 | $100.00 | | $250 | $8.75 | $258.75 | $250.00 |

The effective rate drops as your top-up grows — from 9.5% at the $5 minimum down to 3.5% flat at $100 and above. Larger top-ups are meaningfully cheaper per dollar of credit.

Same fee, every payment method

The service fee is the same whether you pay by credit card, debit card, ACH transfer, or a digital wallet. We do not charge a higher fee for foreign cards. We do not break out an FX or international card line item. The fee structure is uniform.

Top-up chips and custom amounts

The dashboard offers four preset top-up amounts:

$10 — quick small refill.
$25 — modest top-up for development traffic.
$100 — most common top-up; lands at the flat 3.5% rate.
$250 — for production accounts running consistent volume.

You can also enter a custom amount. The minimum top-up is $5.

The Stripe checkout flow shows you the credit amount, the service fee, and the all-in total before you click pay — there is no "surprise at the end."

Trial credit

When you create your account and attach a payment method, we grant you $5 in trial credit. You don't pay anything for it. We don't charge your card.

The trial credit:

Is spendable on /v1/messages calls only (not /v1/chat/completions at V1).
Spends down on real usage at the standard 9%-markup pricing.
Has no expiration.
Is granted exactly once per account.

After the trial credit is exhausted, you spend against your paid balance. If your paid balance is also empty, requests return a 402 Payment Required until you top up.

Auto-topup

You can configure auto-topup in the dashboard:

Threshold: when your balance drops below this amount, we automatically run a top-up.
Amount: the size of the auto-top-up.

Auto-topup charges your default payment method using the same service-fee structure as a manual top-up. If the auto-topup fails (declined card, expired card), we mark your account and send you an email — but we do not block your in-flight requests. You can spend your remaining balance down to zero. Only when the balance actually reaches zero will requests start returning 402 Payment Required.

To clear an auto-topup failure: update your payment method in the dashboard or via the Stripe Customer Portal. The next time your balance crosses the threshold, auto-topup re-arms.

Refunds and chargebacks

We don't issue refunds on completed top-ups — credit on your account is non-refundable. You can hold unused credit indefinitely; it does not expire and is portable across keys.

If your card is charged in error, contact support at support@unnma.ai with the transaction details and we will work it out.

See your bill

The Billing screen in the dashboard shows:

Your current paid balance and any remaining trial credit.
A history of top-ups, auto-topups, and welcome credits.
A daily usage rollup — what you spent, by day, by model.
Receipts for every transaction (also available via the Stripe Customer Portal).

Authentication — managing keys for environments and services.
Rate limits — per-key rate limits and daily spend caps.
Error codes — what 402 insufficient_funds actually looks like.