Skip to main content
AI companies pioneered usage-based pricing with tokens, credits, and compute-based billing. This model aligns cost with value: customers pay proportionally to their consumption.

​
Common AI Pricing Models

1

Create usage-based products

Navigate to Products → Add Product
  • Create “API Usage” as a metered product
  • Price per 1,000 or 1,000,000 tokens/units
  • Consider separate pricing for input vs. output tokens
2

Set up billable metrics

Navigate to Billable Metrics → Create Metric
  • Metric: “Input Tokens” (sum of input_tokens property)
  • Metric: “Output Tokens” (sum of output_tokens property)
  • Metric: “API Calls” (count of requests)
3

Implement credit system (optional)

Use Alguna’s credits/wallet feature:
  • Customers pre-purchase credits
  • Usage depletes credits in real-time
  • Auto top-up when balance is low
  • Overage billing when credits exhausted
4

Create tiered pricing for volume

Reward high-volume customers:
  • First 1M tokens: $3.00 per 1M
  • 1M - 10M tokens: $2.50 per 1M
  • 10M+ tokens: $2.00 per 1M
5

Set up real-time usage tracking

  • Ingest usage events via API
  • Provide customers with usage dashboards in portal
  • Configure alerts at 50%, 80%, 100% of limits

​
Billing Flow

Post-paid usage is logged and rated at the end of the period. Credit-backed usage can instead be authorized before the work runs — see below.

​
Credit System vs. Post-Paid

​
Gate Expensive Work Before It Runs

Inference calls, long-running jobs and GPU work cost you money the moment they start. Rating usage after the fact means a customer with an empty balance can still run up cost you cannot recover. Alguna supports authorizing work up front, as a two-phase flow:
1

Reserve

POST /credits/check with reserve prices the work and opens a hold, returning a reservation_id. You describe the work as the events it will produce — you never send a credit amount, and pricing stays server-side. An Idempotency-Key is required whenever reserve is set.
2

Settle

POST /credits/track with the reservation_id and the events that actually happened charges the real cost and closes the hold. If the work never ran, close the reservation with no events and nothing is charged.
This is the difference between discovering an overdraft on the invoice and refusing the request that would have caused it.

Realtime credit gate

Full reserve/settle reference: request and response shapes, idempotency, drawing without a reservation, reading a reservation, and releasing abandoned work.

​
Key Features to Configure

Billable Metrics

Define what usage events to track and how to aggregate them.

Credits & Wallets

Set up pre-paid credit systems with auto top-up.

Usage Metering

Ingest usage events and see real-time consumption.

Credit Consumption

Consumption order, ledger entries, and the realtime reserve/settle gate.

Customer Portal

Let customers view their usage and manage billing.

​
API Reference

Ingest Events

Send usage events to Alguna.

Credit Grants API

Manage customer credit balances.

Webhooks

Get notified of billing events.

Customers API

Create and manage customer accounts.

Self-Serve (PLG)

Most AI companies start with self-serve checkout.

Developers

API integration for usage metering.

Hybrid Motion

Land self-serve, expand with sales for enterprise.

← Back to Quick Starts

See all quick start guides