Pricing
Compute at cost + 1%. That’s the whole model.
No seats, no tiers, no “contact us” for the real number. One ledger, every fact a row.
| Fact | Self-serve | Enterprise |
|---|---|---|
| Platform fee | compute cost + 1% pass-through | $5,000/yr flat |
| Compute markup | +1% | +0% — compute at cost |
| Billing | prepaid credits | annual invoice |
| Free credit | $1.00 on email verification, no card | pilot on your own traffic |
| Minimum top-up | $5.00 | — |
| Routing | cheapest right-sized model, automatic escalation | same router, tunable per-route quality bars |
| Fine-tuned models | — | included — tuned on your traffic |
| Deployment | shared cloud | shared cloud or VPC option |
| Receipts | per-response x-vynaris-* headers + CSV export | same, plus monthly signed statement |
| Support | email, public issue tracker | direct line to the founder |
| Start | Get your API key → | See the math → |
Compute billed at provider list price. Every request’s served model and cost appear on the response and in your exportable ledger.
Billing rules we hold ourselves to
Three pledges, in writing.
Credits never expire
A deposit is your money until you spend it. No expiry dates, no “inactive account” forfeiture, ever.
We never inflate the savings number
Savings are computed per request, against the model you actually asked for, at list price — and the raw ledger is exportable as CSV so you can recompute it.
No auto-enrolled auto-top-up
Auto-top-up exists, but it is off until you turn it on, with a cap you set. Running out of credit stops requests; it never starts charges.