# Usage per customer

> If your app serves many customers, KONG API AI records usage per customer and lets you set a separate limit for each one.

Source: https://kongapiai.com/en/docs/customers

## When to use it

You build an app and have customers of your own. You want to know how many tokens each customer uses, and to stop a customer who goes over the agreed amount.

Customers and agents count as keys of your plan, not as members.

## Easiest: one key per customer

In the console, create a key with the customer selected. Every request of that key is attributed to that customer. Nothing else to send.

## One key for many customers

1. In the console, open **Customers**, open a customer and choose **Generate Token**. The token is signed and valid for 15 minutes.
2. Send the token in the `X-GW-Customer` header of each request.

```bash
curl https://kongapiai.com/v1/chat/completions \
  -H "Authorization: Bearer $KONG_KEY" \
  -H "X-GW-Customer: $CUSTOMER_TOKEN" \
  -H "Content-Type: application/json" \
  -d '{"model": "MODEL_ID", "messages": [{"role": "user", "content": "Hello"}]}'
```

> A customer token expires after 15 minutes. Do not leave one in a client-side app for long.

## A limit for a customer

Each customer can have a monthly limit. At the limit, that customer's requests get `402`; other customers keep running.

The limit counts only what is paid with KONG API AI credits. Usage on your own sources is not capped. See [Limits and routing](https://kongapiai.com/en/docs/limits).

## Turn a customer off

When a customer is turned off in the console, their requests get `403` with `customer_disabled`.
