/Docs
Search docs or ask a question⌘K
DocsAgent
Changelog
API API Reference Rate Limits

Rate Limits

How rate limiting and usage limits work across the Syntext API.

4 min read

Syntext applies two kinds of limits: request rate limits (short-window, per-client) and usage limits (monthly, per-organization, based on your plan).

Request Rate Limits

When you exceed a rate limit, the API returns 429:

json
{
"error": {
"code": "RATE_LIMITED",
"message": "Too many requests. Please wait before asking another question.",
"retryAfterSeconds": 30
}
}

Wait retryAfterSeconds before retrying.

AI Assistant Visitor Limits

The public chat endpoint is rate-limited per visitor to protect your AI query quota from abuse. The limit is a per-project setting (aiRateLimit); it isn't editable in the dashboard. When a visitor exceeds it, the widget shows a cooldown message.

Agent mode action limits

The public agent mode actions — report a docs gap, share a conversation, draft and send to support — have fixed limits, counted per visitor, per client IP and per project:

ActionPer visitorPer client IPPer projectPer organization
Report a docs gap5 an hour20 an hour200 an hour—
Share a conversation20 an hour60 an hour500 an hour—
Draft a support handoff5 an hour15 an hour100 an hour—
Send to support3 a day10 a day50 a day200 a day
Run an API request20 per 10 minutes60 per 10 minutes1000 an hour2000 an hour
Gated pages notice check30 an hour60 an hour2000 an hour, and 10,000 a day—
•
The per-IP limit applies to requests that come through your doc site (/_chat/...). Requests made straight to the API are limited per visitor and per project only.
•
Runs only come through the doc site (direct calls get 403), and every limit is enforced; if the limits can't be checked, the run is refused. On top of the hourly limits, at most 5 runs per project and 20 per organization are in flight at once. There is no global run budget.
•
Gated pages notice checks are made for every eligible question, whether or not anything matched. Over a limit, or if the limits can't be checked, the notice just doesn't appear — the answer is unaffected and there's no 429.
•
A request is counted only once the conversation is found to belong to the visitor, so requests for other conversations never use up a project's allowance.
•
If the limits can't be checked, the two support actions are refused rather than let through; the others are allowed.
•
Stopping sharing and reading a shared link have no action limits.

Over a limit, the response is 429 with a Retry-After header:

json
{
"error": {
"code": "RATE_LIMITED",
"message": "Too many requests. Please try again later.",
"retryAfterSeconds": 1840
}
}

Editor limits

The web editor's two most expensive requests are limited per user, per project, per minute:

RequestLimit
Editor chat — messages to the editor's Assistant20 a minute
Editor symbols — the @stx symbol index30 a minute

Over a limit, the response is 429 with the wait in details.retryAfterSeconds:

json
{
"error": {
"code": "rate_limited",
"message": "Too many requests. Wait a moment and try again.",
"details": { "retryAfterSeconds": 42, "traceId": "4f7c2a9e1b3d" }
}
}

Account limits

RequestLimitOver the limit
Change Email5 an hour per account429 rate_limited, Retry-After: 3600
Unsubscribe and Get Unsubscribe Link30 a minute per client IP, across both429 rate_limited

Monthly Usage Limits

Plan-based limits are enforced per organization and reset monthly:

Metered resourceApplies to
BuildsTrigger Build, Deploy, Roll Back, git-push builds, PR previews
AI queriesAI assistant questions across all your doc sites

When a limit is reached the API returns 402 with code USAGE_LIMIT_EXCEEDED (earlier releases returned 403, so match on the code). Creating a project beyond the plan's projects limit returns the same error:

json
{
"error": {
"code": "USAGE_LIMIT_EXCEEDED",
"message": "Monthly AI query limit reached (1000/1000). Upgrade your plan for more AI queries.",
"current": 1000,
"limit": 1000
}
}

Current usage is visible in the dashboard under Billing in the account sidebar.

Checking current usage

Get Usage returns this month's usage next to the organization's limits.

Best Practices

•
Poll builds at ≥5s intervals — or use build status webhooks instead of polling.
•
Batch content updates — use the batch save endpoint instead of one request per file.
•
Cache chat/config — widget configuration rarely changes.
•
Respect retryAfterSeconds — retrying earlier only extends the cooldown.
Note

Analytics beacon endpoints (/analytics/pageview, /analytics/search, /analytics/chat) are designed for high-volume traffic from doc sites and return 202 immediately.

Was this page helpful? Yes No
Previous
Product
Resources
Company
© 2026 SyntextBuilt with Syntext