Rate limits
How requests are throttled per client IP, what a throttled client gets back (HTTP 429, rate-limit headers), and where object limits live.
To keep the service responsive and to blunt abuse, the platform limits how many requests a single client may make in a short window. This page explains the request rate limiting that applies to sign-in, the account area, and the REST API, what a throttled client receives, and how it differs from the account-protection features documented elsewhere.
Rate limiting here is about request volume. It is separate from the login-surface defenses (per-IP rate limiting, brute-force protection, account lockout) covered under Threat protection, and from the per-tier object limits (how many users, applications, and so on you may create) documented under Plans and add-ons.
How request throttling works
- Requests are counted per client IP address, not per user or per session, so an unauthenticated caller is throttled just the same. An IPv6 client is counted by its /64 prefix, since one connection usually holds the whole prefix.
- The count is shared across the whole deployment. Because every node reads the same counter, a client cannot get more requests through by reaching a different node.
- Sign-in, account, and REST API traffic are throttled. When a client exceeds the allowed number of requests in the window, further requests are rejected with HTTP 429 (Too Many Requests) until the window rolls over. The limit is self-expiring: normal traffic resumes automatically once the client slows down.
- A separate platform-wide protection guards against high-volume floods (denial-of-service attempts) on top of the per-IP window.
- Emails that an unauthenticated caller triggers to an address it enters (sign-up verification, its resend, the password-reset link) are additionally limited to 5 per hour per recipient address, whatever IP the requests come from. Beyond that, a new sign-up or reset request for the address is refused, and a resend sends nothing, until the hour has passed.
What a throttled client receives
A throttled REST API or sign-in request returns 429 together with headers that tell the client when to retry:
| Header | Meaning |
|---|---|
Retry-After | Seconds to wait before retrying. |
X-RateLimit-Limit | The number of requests allowed in the window. |
X-RateLimit-Remaining | Requests still allowed in the current window. |
X-RateLimit-Reset | When the window resets, as a Unix timestamp (seconds). |
Build clients to honor these headers. On a 429, wait for Retry-After and retry, rather than looping immediately. Reading X-RateLimit-Remaining lets a well-behaved client slow down before it is throttled. Treat the headers as the authoritative limit; the exact numbers can change, and a client that reads them adapts automatically.
Machine-to-machine token requests
Tokens obtained with the client-credentials grant are limited separately, per client rather than per IP address. The token endpoint also serves interactive sign-ins, and one misbehaving service must not throttle the logins of everyone sharing its network. The allowance depends on your subscription:
| Tier | Token requests per minute, per client |
|---|---|
| Trial | 30 |
| Free | 60 |
| Essential | 1,200 |
| Advanced | 6,000 |
| Pro | 30,000 |
A client over its limit receives 429 with a Retry-After header and the OAuth error too_many_requests. Interactive sign-ins are never affected by this limit.
A service that caches its access token and reuses it until shortly before expiry needs roughly one token request per hour and stays far below every tier's allowance. Reaching the limit almost always means the client requests a fresh token for every API call. Fix the caching rather than asking for a higher limit. The entry tiers additionally have a monthly token budget; both limits are documented under APIs.
What is not rate-limited this way
- Account lockout and IP blocks from repeated failed logins are a different mechanism, configured under Security and cleared under Monitoring → Rate limits and blocked IPs. Those block a specific IP or lock a specific account for a longer, configurable period; request throttling only delays a burst of requests.
- Object and sizing limits (maximum users, applications, connections, policies, network zones, admin team members) are governed by your subscription and license, not by request rate. See Plans and add-ons.
Related
- Threat protection, covering per-IP rate limiting, brute-force protection, breached-password detection, and the bot detection on the login surface.
- Rate limits and blocked IPs, for reviewing and releasing blocked clients.
- Plans and add-ons, for per-tier object limits.
- APIs, for obtaining and using tokens for the REST API.