Guide
Rate limits
Per-key caps and the Retry-After header.
Limits are per key, per minute, sized so real products never notice them:
| Surface | Limit |
|---|---|
| Chat completions | 60 / min |
| Image generation | 20 / min |
| Video, speech, transcription | 6 / min each |
On 429, wait exactly Retry-After
Hitting a limit returns HTTP 429 with a Retry-After header (seconds). The window slides, so backing off exactly that long always clears it — retrying sooner just resets your queue position.
Need more for a launch? [email protected].