Rate limits exist so one runaway loop cannot degrade the runtime for everyone, including you. They are generous for normal use and visible before you hit them.
The limits
- Starter — 60 requests/min, 5 concurrent runs.
- Pro — 600 requests/min, 25 concurrent runs.
- Team — 3,000 requests/min, 100 concurrent runs, pooled per workspace.
- Enterprise — set per contract. Tell us your shape and we provision for it.
What hitting a limit looks like
You get a 429 with a Retry-After header that means it. The SDKs retry with exponential backoff automatically; if you call the API directly, honor the header instead of hammering — retries that ignore it count against the next window.
HTTP/1.1 429 Too Many Requests
Retry-After: 12
x-mynd-limit: 600
x-mynd-remaining: 0
x-mynd-reset: 2026-06-10T09:14:30ZPractical advice
Check x-mynd-remaining and pace yourself before the wall, batch where the API offers batch endpoints, and give bursty jobs their own workspace so they cannot starve interactive use. If you have a legitimate sustained need above your tier, ask — raising a limit for a real workload is a five-minute decision on our side.