Every modern monetization model, in one engine.

The pricing question is no longer which model — it is how do I combine the right ones for the right segment? Below is a practical reference for the six dominant models in API and AI monetization, what each is good for, and where each breaks without carrier-grade rating underneath it.

Pay-as-you-go

The original usage model. Customers pay for exactly what they consume — per call, per token, per event, per message. Simple to communicate, easy to start with, but unforgiving on margin if pricing is wrong, and exposed to bill shock without proper visibility and alerts.

Best for: developer-led adoption, low-stakes consumption, infrastructure APIs. Pitfall without carrier-grade rating: lost revenue from un-rated events, mismatched units, or quiet schema drift.

Tiered subscription

A flat price for a usage bucket — up to 1M calls / month for $X. Predictable for buyers, easy to forecast, and the foundation of most enterprise plans. Tiers force a packaging conversation: where do the breakpoints sit, what is included, what counts as overage?

Best for: committed customers, finance-friendly buyers, predictable workloads. Pitfall without carrier-grade rating: wrong-tier customers and unbilled overages are the leakage vectors here.

Hybrid (subscription + usage)

The dominant modern shape. A platform fee plus consumption — sometimes with included allowance, often with discounts at volume, frequently with prepaid commitments and true-up at quarter-end. Hybrid is the structure most B2B software is moving toward.

Best for: most API-first SaaS, AI products, IoT platforms. Pitfall without carrier-grade rating: allowances, overages, commitments, and credits compound complexity. This is where naive billing systems break.

Prepaid credits

Credits abstract the underlying unit. Buy 10,000 credits, spend them however. They simplify the customer experience and absorb pricing changes without renegotiation, but only if the credit balance, expiry, and accrual logic are watertight.

Best for: AI products with multiple model tiers, multi-product platforms, marketplace-style consumption. Pitfall without carrier-grade rating: balance drift, expired credits silently consumed, mismatched ledgers between billing and product.

Outcome-based

Pricing tied to a delivered result — a closed ticket, a successful agent task, a verified document, a recovered dollar. Outcome pricing is gaining ground for AI agents because it aligns price with value rather than with consumption cost.

Best for: AI agents, automation, services-as-software. Pitfall without carrier-grade rating: the outcome event itself must be auditable, deduplicated, and tied to a contractual definition. This is rating, not analytics.

Agent billing

The newest shape. AI agents act, decide, transact. Billing is per task completed, per workflow executed, per decision rendered, sometimes per dollar saved. Agent billing combines event mediation, outcome rating, and customer-readable invoicing.

Best for: AI agent products, services-as-software platforms. Pitfall without carrier-grade rating: agent events are heterogeneous, high-volume, and require provenance — which is exactly what carrier-grade mediation provides.

A quick decision framework

Pricing is no longer one model. It is a combination — and the engine underneath has to handle all of them simultaneously.