End-User Billing: Meter, Limit, and Charge Your Users

If you're running AI features in production, you've already dealt with the first wave of problems: routing, security, observability. But once real users show up, a second problem arrives fast — who's using how much, and who's paying for it.

Today we're launching End-User Billing on Cencori. It gives you per-user metering, rate plan enforcement, markup pricing, and Stripe Connect payouts. All of it works through the same gateway you already route AI traffic through.

The Problem

Most teams building AI products end up cobbling together usage tracking from three or four systems. Token counts from the provider. Cost estimates from a spreadsheet. Rate limiting from a Redis layer they wrote themselves. Billing from Stripe, but with no connection to any of the above.

The result: you either don't charge for AI usage at all (and eat the cost), or you build a fragile internal billing pipeline that breaks whenever you add a model or change pricing.

What We Built

End-user billing is a single system that handles the full loop:

1. Rate Plans — Define tiers with token limits, request limits, cost caps, per-minute burst protection, model restrictions, and your markup on top of provider cost. Create as many plans as you need. Assign a default for new users.

2. Quota Enforcement — Every request through the Cencori gateway is checked against the user's plan limits in real time. Over their daily token cap? Blocked with a 429 and a Retry-After header. On an alert-only plan? Request goes through, you get flagged in the dashboard.

3. Usage Tracking — Every request is metered per user, per day, per month. Token counts, request counts, provider cost, your markup — all tracked automatically. No extra API calls if you use the gateway.

4. Markup Pricing — Set a percentage markup on top of provider cost, a flat fee per request, or both. Different plans can have different margins. The math happens at the gateway level — you see provider cost and your revenue as separate numbers.

5. Invoices — Generate invoices for any billing period. Send them through Stripe Connect or export the data and bill however you want.

6. Stripe Connect Payouts — Connect your Stripe account once (org-level, covers all projects). Generate invoices, send them to your users, collect payment. Your users pay you directly. Cencori never touches the funds.

Two Integration Paths

If you use the Cencori gateway — you're done. Pass user: "user_123" in your AI requests. Billing, limits, metering — all automatic. Zero additional code.

const client = new OpenAI({
  apiKey: "csk_your_cencori_key",
  baseURL: "https://api.cencori.com/v1",
});
 
const response = await client.chat.completions.create({
  model: "gpt-4o",
  messages: [{ role: "user", content: userMessage }],
  user: "user_123",
});

If you call providers directly — two API calls. Check quota before, report usage after.

// Before processing
const quota = await fetch(
  `https://api.cencori.com/api/v1/billing/check-quota?end_user_id=${userId}`,
  { headers: { Authorization: `Bearer ${CENCORI_KEY}` } }
).then(r => r.json());
 
if (!quota.allowed) {
  throw new Error(`Limit reached: ${quota.reason}`);
}
 
// After processing
await fetch("https://api.cencori.com/api/v1/billing/usage-events", {
  method: "POST",
  headers: {
    Authorization: `Bearer ${CENCORI_KEY}`,
    "Content-Type": "application/json",
  },
  body: JSON.stringify({
    end_user_id: userId,
    model: "gpt-4o",
    provider: "openai",
    prompt_tokens: response.usage.prompt_tokens,
    completion_tokens: response.usage.completion_tokens,
    cost_usd: 0.0085,
  }),
});

Environment Scoping

Test-key traffic and production-key traffic are tracked separately. Your QA runs don't blow out a user's production quota. This is automatic — it follows the API key environment.

The Dashboard

Four tabs:

• Configuration — toggle billing on/off, set your default markup, connect Stripe
• End Users — search, filter, block/unblock, change plans, view per-user usage
• Rate Plans — create and manage tiers with full limit and pricing controls
• Revenue — total revenue, provider cost, margin, daily trends, top users

Everything updates in real time as requests flow through the gateway.

Who This Is For

Any team charging for AI usage — or any team that should be but isn't because building the billing pipeline felt like too much work.

SaaS products with AI features. API companies reselling model access. Internal tools that need per-team cost attribution. Enterprise deployments where different departments need different limits.

If you already route traffic through Cencori, this is a configuration change. If you don't, the Usage Events API gets you there in under an hour.

Get Started

• Setup Guide — step-by-step, no code required for the dashboard setup
• API Reference — full technical docs for your engineering team
• Dashboard — enable billing on any project right now

End-user billing is live for all Cencori projects today.