Rate limits for the public API
Limits move from one global cap to per-plan quotas, so a single noisy integration can no longer slow everyone else down.
Limits by plan
Each key gets a per-minute quota and a short burst allowance. Responses carry RateLimit-Remaining and Retry-After so clients can back off on their own.
| Plan | Requests / min | Burst |
|---|---|---|
| Free | 60 | 120 |
| Team | 600 | 1,200 |
| Business | 3,000 | 6,000 |
Rollout
- Send rate-limit headers in shadow mode
- Email Business accounts above 80% of their limit
- Enforce limits on November 30