API reference
Rate limits
Limits keep one caller from starving the others. They are per minute and counted per key and per workspace.
| Scope | Limit |
|---|---|
| Per API key | 60 requests per minute |
| Per workspace (all keys together) | 300 requests per minute |
| Per IP address, before authentication | 120 requests per minute |
| Text per request | 100,000 characters |
When you hit one
The response is 429 rate_limited with a Retry-After header in seconds. Successful responses also carry X-RateLimit-Limit and X-RateLimit-Remaining. Wait for Retry-After, then retry; add jitter if several workers share a key.
Need higher limits? Email [email protected] with your expected volume.