Rate Limits

Segmind rate-limits API requests to keep the platform fair and reliable for everyone. The limit depends on your plan.

Limits by plan

RPM is requests per minute. On Scale, every member and every API key in the team draws from the same 1000 RPM allowance. See segmind.com/pricing for what each plan includes and how to upgrade. For a higher or reserved limit, contact sales about Enterprise.

What counts as a request

The limit counts submissions to model endpoints: POST /v1/{slug} and POST /v2/{slug}. Each submission is one request, whatever the model or the size of the job.

Polling a /v2 request's status_url or response_url does not count against your plan's limit, so you can poll as often as the polling guidance recommends.

There is no per-account cap on concurrent in-flight generations. Long-running video or LLM jobs do not block new submissions. If more /v2 jobs are submitted than workers are free, they wait in QUEUED, and the queue_time field in the result's metrics shows how long.

What happens over the limit

A submission over your limit is rejected with 429 Too Many Requests. Nothing is queued and nothing is charged. Back off and retry: wait a few seconds, then retry with exponential backoff.

Successful /v1 responses carry two headers you can use to pace yourself:

A /v2 job that has already been accepted is not lost if the upstream model is busy: Segmind retries it automatically, and the job stays in QUEUED until it runs.

If you regularly hit your limit, upgrade your plan at segmind.com/pricing. Error codes for other failure modes are listed in the API overview.