On this page

Look up a request

GET /v1/generation returns the metadata of one past request. It never returns prompts or responses.

GET /v1/generation?id=<request id> (API reference) returns the metadata of one request made with your workspace’s keys. Use it to check what a request cost or how long it took, or to give support the facts about one request.

The id is the value of the x-request-id header on the original response.

curl "https://api.router.africa/v1/generation?id=$REQUEST_ID" \
-H "Authorization: Bearer $ROUTER_AFRICA_API_KEY"

import os
import httpx

r = httpx.get(
  "https://api.router.africa/v1/generation",
  params={"id": os.environ["REQUEST_ID"]},
  headers={"Authorization": f"Bearer {os.environ['ROUTER_AFRICA_API_KEY']}"},
)
print(r.status_code, r.json())

const url = new URL("https://api.router.africa/v1/generation");
url.searchParams.set("id", process.env.REQUEST_ID!);

const res = await fetch(url, {
headers: { Authorization: `Bearer ${process.env.ROUTER_AFRICA_API_KEY}` },
});
console.log(res.status, await res.json());

What you get

The response has one data object. This page lists every field; the reference page has the full schema.

Field	Meaning
`id`	The request id.
`created_at`	When the request was made, as a date-time.
`endpoint`	The path that was called, for example `/v1/chat/completions`.
`model`	The model the request used.
`status_code`	The HTTP status the request returned.
`error_code`	The error `reason`, or `null` when the request succeeded.
`latency_ms`	How long the request took, in milliseconds.
`streamed`	Whether the request was streamed.
`input_tokens`	Tokens in the prompt.
`output_tokens`	Tokens in the response.
`cost`	The charge, in credits, as a string with eight decimal places.

{
  "data": {
    "id": "0192f0e0-7a1c-7d3e-9f10-2b6c8a4d5e11",
    "created_at": "2026-11-07T09:30:00Z",
    "endpoint": "/v1/chat/completions",
    "model": "openai/gpt-4.1-mini",
    "status_code": 200,
    "error_code": null,
    "latency_ms": 840,
    "streamed": false,
    "input_tokens": 12,
    "output_tokens": 3,
    "cost": "0.00000240"
  }
}

What you don’t get

No prompts or responses. There is no way to read stored content through the API. Admins read it in the dashboard, and every read is recorded.
No attempt or fallback counts. These are not recorded.

Rules

You can look up only requests from your own workspace. A request from another workspace returns 404, the same as an id that doesn’t exist.
A request is recorded just after its response is sent. If you look it up straight away you can get 404. Wait a moment and try again.