Getting started
Rate limits
Request and concurrency ceilings, with headers on every response.
Two limits apply. A request rate limit on how often you may call, and a concurrency limit on how many jobs may run at once. Reads and writes are counted separately.
| Limit | Default |
|---|---|
| Generation requests | 60 per minute |
| Read requests | 600 per minute |
| Concurrent jobs | 8 |
Exceeding concurrency queues the job rather than rejecting it. Exceeding the request rate returns 429 with a Retry-After header. Higher ceilings are set per account; ask and we will raise them.
X-RateLimit-Limit: 60
X-RateLimit-Remaining: 47
X-RateLimit-Reset: 1789665310
X-Concurrency-Limit: 8
X-Concurrency-Running: 3Back off on 429 rather than retrying immediately. A client that retries tightly stays rate limited; one that honours Retry-After recovers on the first attempt.