Credit-Based Billing: Examples & Companies

What is it

Credit-Based Billing is a pricing model where customers spend a pool of credits across multiple product features, with each feature consuming credits at its own rate.

Credits act as an abstraction layer between the customer and the underlying cost driver. Instead of seeing “0.0001¢ per token” or “$0.0003 per function invocation,” the customer sees “you have 5,000 credits this month.” The vendor can change underlying economics — swap a cheaper model in, change a feature’s credit cost, add a new feature — without renegotiating contracts. The buyer can budget in one number instead of N.

The model is now standard across AI products. Cursor ties $1 of credits to $1 of underlying API cost — the most transparent variant in the category. Vercel uses credits as a flexible spending mechanism across its AI products (v0) while billing infrastructure on per-dimension meters. Other AI tools — Replit, Cline, Codeium — have converged on credit pools as the standard packaging.

How it works

The two key design decisions in any credit system are: (1) what defines a credit, and (2) how credits replenish.

Unit math (Cursor’s transparent model): $20 Pro plan → 20 credits → ~500 GPT-5.1 Codex requests (at $1.25 input / $10 output per million tokens, average request ~10K tokens). Same $20 credits → ~45 Claude 4 Opus requests at $15/$75 per million. A 10x spread across frontier models.

The opaque variant is more common in enterprise SaaS. Notion AI’s “credits” are not tied to any underlying API rate; Salesforce Einstein “credits” likewise. The opacity is deliberate — it lets the vendor change underlying costs without exposing them.

Companies using this

Two companies in the current corpus use credit-based billing: Cursor with its transparent $1-credit-equals-$1-API-cost model, and Vercel which mixes credit-style packaging on v0 with per-dimension billing on infrastructure.

Patterns observed

• Credits unify pricing across multi-product surfaces. Vercel uses credits for v0 (AI UI generation) alongside per-dimension billing for hosting because credits are the right unit when the underlying cost driver is variable per request. The credit-vs-meter choice maps to the cost-driver-variance question.

• Transparent credits create competitive pressure for opaque ones. Cursor’s $1=$1 model made the markup question public for the entire AI coding category. Competitors with opaque credits now have to either disclose their ratio or compete on something other than transparency.

• Credit-pool models pair with overage policies. Both companies in this corpus support opt-in overage — if you exhaust your pool, you can choose to either stop or auto-bill at the posted rate. Neither does silent auto-bill; both default to hard caps.

Counterexamples & variants

The strongest critique of credit-based billing is that it can hide cost-per-feature from the customer. A user spending 80% of their credits on one expensive feature won’t realise it until they look at the per-feature breakdown — if the vendor even shows one. Cursor added per-model spending visibility after the June 2025 incident; competitors with no such breakdown remain bill-shock-prone.

Variant worth noting: credits-without-pool. Some vendors charge per-credit without including a baseline pool — every credit is paid usage. This is functionally identical to pure-usage pricing with a credit denomination — it’s a marketing choice, not a structural one. The corpus classifies these as pure-usage, not credits.

What this means for buyers vs vendors

For buyers

Demand transparency on credit consumption rates per feature. A pricing page that lists credit cost per feature is structurally honest; one that shows only “credits per month included” is structurally opaque. In procurement, ask for the worst-case-per-feature credit rate and run it against your projected usage.

For vendors

Credits are the right packaging when (1) you have multiple features with different cost drivers, (2) underlying costs may shift over the contract period, or (3) the buyer’s mental model is “budget per month” rather than “cents per request.” If those don’t hold, credits add complexity without buyer benefit — stick to direct per-unit billing on the dominant cost driver. See our usage invoicing guide for the surrounding infrastructure choices.