Rate limits

Four limits apply, checked in this order:

Limit	Default	Applies to
Per minute, per IP	300 requests/min	Every request, before your key is checked
Per minute, per key	60 requests/min	Every authenticated request
Per day, per key	50,000 requests/day, reset at 00:00 UTC	Every authenticated request
Per minute, per key, writes	10 requests/min	`POST`, `PUT` and `PATCH` calls. Writes also count toward the per-minute limit.

Limits are approximate. Treat the headers as guidance, not an exact count.

Headers

Every response from an authenticated call carries:

RateLimit-Limit: 60
RateLimit-Remaining: 57
RateLimit-Reset: 42

RateLimit-Reset is the number of seconds until the current window resets. These headers are set on every response, not only on a rejected one. Use them to back off before you are throttled. A rejected request also carries:

Retry-After: 42

When you are throttled

Code	HTTP	Meaning
`rate_limited`	429	The per-minute limit or the write limit is spent.
`daily_quota_exceeded`	429	The 50,000/day quota is spent. It resets at 00:00 UTC.

Both carry retry_after_seconds in the error body, matching the Retry-After header.

{
  "error": {
    "type": "rate_limit_error",
    "code": "rate_limited",
    "message": "Too many requests. Slow down and retry after the interval in `Retry-After`.",
    "doc_url": "https://docs.gpuoutlet.ai/api/errors#rate_limited",
    "request_id": "a1b2c3d4-...",
    "retry_after_seconds": 42
  }
}

Retrying

Read Retry-After or retry_after_seconds, and wait at least that long before your next call. A script that retries on its own should back off with jitter, not retry at a fixed interval. Retries that all arrive the moment the window resets get throttled again. Writes have a smaller budget than reads. It stops a runaway script from filing many requests or declines before anyone notices. Build polling around the per-minute limit. Keep the write budget for occasional changes.

Idempotency

Safe retries for writes, independent of rate limiting.

Errors

Every error code this API returns.

Get started

Conventions

Workflows

Endpoints

Headers

When you are throttled

Retrying

Idempotency

Errors

​Headers

​When you are throttled

​Retrying

​Related

Idempotency

Errors

Headers

When you are throttled

Retrying

Related