Design billing that matches AI usage and preserves your margins. You can change pricing and rules in the dashboard at any time without touching your app code. You can use one method or combine them.

​
1. Deposit credits via Deposit API (Default mode only)

What it does: Credit a user’s app‑scoped wallet with an arbitrary amount. Use cases
  • Integrate any payment system (PSP, crypto, app store) — after you confirm payment, deposit credits.
  • Grant trial credits to new users or cohorts.
  • Add credits at the start of a billing cycle or for goodwill/refunds.
Notes
  • Appears as a deposit in the ledger (developer‑initiated).
  • Works only in Default mode (virtual, app‑scoped balance).
  • Pair with the automated Stripe restricted key setup for a zero‑code top‑up flow.

​
2. One‑off charges via Charge API (Both modes)

What it does: Create a manual, one‑time charge independent of token usage. Use cases
  • Bill for tool usage (e.g., MCP tool run, file conversion, image upscaling).
  • Charge for feature access (e.g., unlock premium mode or export).
  • Apply post‑processing fees (e.g., retrieval/storage, long‑running jobs).
Notes
  • Appears as a charge in the ledger.
  • If funds/authorization are missing, the API returns a renderable message (auth/top‑up link); show it as a normal assistant reply.
  • Use an Idempotency‑Key to avoid duplicates; include metadata for reconciliation.

​
3. Automatic usage billing via the Proxy (Both modes)

What it does: The proxy meters tokens and computes the charge from the selected model price + your markup. It writes the charge to the ledger and returns the model response (or an auth/top‑up message) — no custom billing code required. Setup
  • Connect a provider: BYOK (Default required; Shared optional) or the built‑in provider (Shared).
  • Set per‑model prices and an optional markup % in the dashboard.
Behavior
  • Charge = usage (prompt + completion tokens) × model price ± markup.
  • Per‑model pricing lets users pick price/performance while you keep margins predictable.
  • Changes take effect without code changes.

​
4. Per‑request & hybrid pricing (Coming soon)

Configure fixed per‑request fees, or hybrid pricing: minimum fee per request + per‑token usage. Useful for short prompts, tools/actions, and stabilizing margins across mixed workloads.

​
Recommendations & guardrails

  • Prefer per‑model dynamic pricing and set a minimum per‑request fee to cover short prompts.
  • Add max tokens and spend caps to prevent runaway costs.
  • Keep pricing rules in the dashboard; don’t hardcode rates in your app.
  • For UI display, read prices/models from the separate API Reference (Models).