Choose how you pay
One API for every major model. Inference is billed at the provider's own rates — we add no markup and no fees.
Pay as you go
Pay as you go
Pay for what you use
- No top-up fees, no markup on inference
- Compare every provider's price for the same model
- Token-level usage detail; cost returned in the response body
Best for
Production workloads, commercial products, teams that need cost attribution
Subscription
Subscription
A flat monthly fee
- Flat monthly fee, predictable cost
- Quota-based; falls back to pay as you go when the quota runs out
- Built for individual development and learning
Best for
Individual development, learning, non-production exploration
Detailed comparison
| Feature | Pay as you go | SubscriptionComing soon |
|---|---|---|
| Billing model | Billed on actual usage | Flat monthly fee + quota |
| Best for | Production, commercial products | Individual development, learning |
| Cost predictability | Tracks usage | High (flat fee) |
| Platform markup | None — provider list price | To be decided |
| Top-up fee | None | Not applicable |
| Provider comparison | Yes — pin a provider or use auto routing | To be decided |
| Usage detail | Token level, including cache and reasoning tokens | To be decided |
| Refunds | Unused credit refundable to the original method — see the refund policy | To be decided |