How It Works
The Metering Infrastructure
for AI Products
Two API calls in your request handler. One API call at billing time. That's the entire integration — and you never build metering infrastructure again.
Check · Call · Record · Bill
Check limits, make your AI call, record usage. We handle the rest.
Step 1
Check
Verify customer is within limits before the call
Step 2
Call
Call OpenAI, Anthropic, or any LLM provider
Step 3
Record
Log tokens and costs per customer
Step 4
Bill
Pull usage data via API for billing
Simple Check-Record Pattern
Two API calls per request. Full control. Works with any LLM.
app/api/chat/route.ts
import Limitry from "limitry";
import OpenAI from "openai";
const limitry = new Limitry({ apiKey: process.env.LIMITRY_API_KEY });
const openai = new OpenAI();
export async function POST(req: Request) {
const { customerId, prompt } = await req.json();
// 1. Check — is the customer within limits?
const check = await limitry.limits.check({
customerId,
meterId: "llm-tokens",
});
if (!check.allowed) {
return Response.json(
{ error: "Usage limit reached", resetsAt: check.resetsAt },
{ status: 429 },
);
}
// 2. Call — make the AI request as usual
const completion = await openai.chat.completions.create({
model: "gpt-4o",
messages: [{ role: "user", content: prompt }],
});
// 3. Record — tell Limitry what happened
await limitry.events.record({
customerId,
eventType: "llm.completion",
value: completion.usage?.total_tokens ?? 0,
dimensions: { model: "gpt-4o", team: "engineering" },
});
return Response.json({
text: completion.choices[0].message.content,
});
}