Operations
Rate limits and concurrency
Requests per second, concurrent conversions and how to retry.
Two limits apply to /v1/convert and /v1/render: requests per second, with a burst, and concurrent conversions. Their values depend on your plan; see Plans and limits.
Both apply per project: every key of a project shares them.
Headers
| Header | Content |
|---|---|
X-RateLimit-Limit | Requests allowed per second. Absent when the plan has no rate limit. |
X-RateLimit-Remaining | Requests left before the rate limit applies; can exceed X-RateLimit-Limit after a quiet period. Absent when the plan has no rate limit. |
Retry-After | On a 429, seconds until a retry can succeed. |
429 responses
A 429 carries rate_limit_exceeded, concurrency_limit_exceeded or failed_render_limit_reached; see Error codes.
Client guidance
- On a 429, wait
Retry-Afterseconds before retrying. - Cap the requests you have in flight at your plan's concurrent conversions, across every key and process.