concurrency_limit_exceeded — concurrent order limit reached
What happened
The order was not created. The body has no extra fields, and there is no Retry-After header.
Why
The limit caps how many of your orders run at the same time. The request rate and the generation volume do not matter here. A batch of hundreds of orders sent at once hits exactly this limit, even when the quotas are free.
What to do right now
Wait for your active orders to finish. That takes minutes, not until the end of the billing period: as soon as one of them finishes or stops, a retry goes through. Pauses between requests do not help: a slot is freed by a finished order. Cap the number of concurrent orders with a queue on the client side. Retrying with a growing delay is the wrong tool here.
Which request to send next
GET /usage: the entry with limit_type: "concurrent" shows the ceiling and the current number of active orders. Its period_start, period_end, and resets_in are null because this limit has no period. Slots are held by orders in the created, pending, and in_progress statuses. Find them with GET /orders?status=…. A WebSocket event is a more convenient way to catch a freed slot than polling.
Where to go next
This page for AI
API version 1.0.0