Pay for an AI API with Crypto (USDT)

Pay for an AI API with crypto (USDT) by funding your Aigenway balance through Cryptomus instead of a card — send USDT, BTC, ETH or another supported coin to an invoice, and the deposit lands as prepaid USD on your account automatically. No card number, no billing address, no bank in the loop. From there every model — image, video, text — draws against the same balance through one task contract.

How a crypto top-up becomes USD balance

Top-ups live under /top-up/new: pick a USD amount between $5 and $10,000, choose Cryptomus as the provider, and the dashboard creates a payment invoice priced in that USD amount. Cryptomus hands back a hosted payment page and a QR code with the coin-specific address and the exact crypto amount to send — pay with USDT on whichever network Cryptomus quotes, or switch to BTC or ETH if that's what's in your wallet.

Once the chain confirms the payment, a webhook from Cryptomus credits your Aigenway balance for the invoiced USD amount — no manual reconciliation, no support ticket to "find" your deposit. The invoice page shows the same status your dashboard shows: pending, paid, or expired if you didn't send funds in time. A paid invoice's USD lands on the exact ledger /v1/generations and /v1/estimate read from — a crypto top-up isn't a separate wallet you have to track alongside a card-funded one.

Spend it: a generation call against the balance you just funded

Once the invoice shows paid, the balance is live immediately — call any model with the same request shape regardless of how you funded the account:

curl -X POST https://api.aigenway.com/v1/generations \
  -H "Authorization: Bearer sk_live_..." \
  -H "Idempotency-Key: $(uuidgen)" \
  -H "Content-Type: application/json" \
  -d '{
    "provider": "google_banana",
    "model": "gemini-3.1-flash-image-preview",
    "method": "text_to_image",
    "params": { "prompt": "a neon-lit night market, rain on the pavement" }
  }'

Idempotency-Key matters just as much on a crypto-funded balance as a card-funded one: resend the same key after a dropped connection and you get the same task back, not a second charge against the USDT you just converted. Don't know the exact model codes you want to call? List what's live on the account first:

curl "https://api.aigenway.com/v1/methods?provider=google_banana" \
  -H "Authorization: Bearer sk_live_..."

One ledger, no matter which coin funded it

A Cryptomus deposit doesn't create a separate crypto wallet inside your account — it's converted and credited as the same USD balance a card top-up would produce, in the same nano-dollar units /v1/estimate and every charge already use. That means:

• /v1/estimate works the same. Price a request against your live rates before committing, and the response's enough_balance check reads the crypto-funded balance exactly like any other.
• Bonus coupons still apply. A DISCOUNT_TOPUP coupon code works at invoice-creation time the same way it would on a card top-up — the bonus lands in USD on top of the deposited amount, not in a coin-specific side ledger.
• Billing history is one list. /top-up shows every invoice — pending, paid, expired — regardless of which coin funded it, so reconciling spend doesn't mean checking a crypto exchange's history alongside a card statement.

Why crypto-first billing matters for an AI API

Most providers gate API access behind a card: a subscription tier, a stored payment method, sometimes a company card to get past fraud checks. That's a real barrier if you're a solo developer without a business card, operating from a region a provider's card processor flags, or simply don't want to hand a generation API recurring access to a card number for a workload you might use once this month.

A prepaid crypto top-up sidesteps all of that — you send exactly what you intend to spend, once, and the balance draws down per request with no recurring authorization sitting on file. Combined with one task contract (provider/model/method/params) across every provider, that means the same funding flow — and the same API — works whether you're calling an image model from a laptop in a country with limited card rails or wiring a production pipeline from a company that'd rather not put a generation API on a corporate card at all.