Skip to content
RelayRouter
On this page

Account

Usage analytics

Aggregate spend, request counts, tokens, and a per-model breakdown.

GET/usage?days=7JWT or API key

days is clamped to 1–90 and defaults to 7. The response powers the dashboard charts and the usage page.

Response

{
  "days": 7,
  "balance_micro": 1522976,
  "balance_usd": 1.522976,
  "totals": {
    "requests": 331,
    "prompt_tokens": 25000000,
    "completion_tokens": 120000,
    "cost_micro": 1006802,
    "cost_usd": 1.006802
  },
  "per_day": [
    {
      "day": 1792195200,
      "requests": 12,
      "prompt_tokens": 8000,
      "completion_tokens": 300,
      "cost_micro": 14000
    }
  ],
  "by_model": [
    {
      "model": "deepseek-v4.1-flash-1017",
      "requests": 320,
      "prompt_tokens": 24000000,
      "completion_tokens": 100000,
      "cost_micro": 980000
    }
  ]
}
  • day is a unix timestamp at UTC midnight.
  • cost_micro already includes the network fee.
  • by_model is ordered by cost and limited to the top 50 models.
  • per_day only includes days with usage; zero-filled rows are a client concern.

What is not available

There is no per-request log endpoint and no per-key breakdown yet. Use the aggregates for charts and the ops CLI for request-level debugging.

Fetching usage

const usage = await fetch("https://api.relayrouter.org/usage?days=30", {
  headers: { authorization: `Bearer ${jwt}` },
}).then((res) => res.json());


console.log(usage.totals.cost_usd);
PreviousJanNextBilling & deposits