Sentry rate limits every API request per caller and per endpoint, with a requests-per-second limit on a fixed window and a separate limit on concurrent requests. Each endpoint has its own maximum, so there's no single number to quote. Response headers show your limit and what's left, and Sentry recommends webhooks instead of polling.
If you've searched for one Sentry API rate limit and come up empty, that's because it doesn't exist. The limits are set endpoint by endpoint, and the only reliable way to know yours is to read the headers that come back with every response.
Two limiters, applied per caller and endpoint
Sentry rate limits every API request to prevent abuse and resource overuse. The limit applies to each unique combination of caller and endpoint, so heavy use of one endpoint doesn't eat into another.
Two separate limits run at once. The first restricts how often you make requests, in requests per second. It uses a fixed window: requests are counted into time buckets, and once you go over the allowance inside a window, the request is rejected. The second restricts how many requests you can have in progress at the same moment; if too many are running, new ones are turned away.
Each endpoint has its own maximum number of requests and its own window size. A busy listing endpoint and a rarely used settings endpoint won't share the same numbers.
The headers that tell you your limit
| Header | What it tells you |
|---|---|
| X-Sentry-Rate-Limit-Limit | The maximum requests allowed in the current window |
| X-Sentry-Rate-Limit-Remaining | Requests left in the current window |
| X-Sentry-Rate-Limit-Reset | When the next window begins, in UTC seconds from epoch |
| X-Sentry-Rate-Limit-ConcurrentLimit | The maximum concurrent requests |
| X-Sentry-Rate-Limit-ConcurrentRemaining | Concurrent requests still available |
Every response includes these. Log them for each endpoint your integration calls during development, and you'll have the real numbers for your account rather than a guess.
More tokens won't help
Here's the detail that stops the obvious workaround. According to Sentry Docs on rate limits, the limiter looks at the caller's identity, not the token or cookie. Generating extra tokens for the same caller doesn't give you more allowance; every request still counts against the same identity. So if one script is hitting the limit, splitting it across five tokens just spreads the same problem around.
Building a client that stays under
- Read X-Sentry-Rate-Limit-Remaining on every response and slow down as it nears zero.
- When you run out, wait until the time in X-Sentry-Rate-Limit-Reset before sending more to that endpoint.
- Cap how many requests your code runs in parallel, using X-Sentry-Rate-Limit-ConcurrentRemaining as the guide.
- Track limits per endpoint, since each one has its own window and maximum.
The reset header is given in UTC seconds from epoch, so convert it to your server's clock before you schedule the next request.
For scheduled jobs, such as a nightly export of issues, spread the calls out rather than firing them all at the start of a window. Because the window is fixed, a burst at the start can use the whole allowance in a second and leave the job waiting for the reset.
Use webhooks instead of polling
Sentry says plainly that polling the API for updates is likely to trigger rate limiting quickly, and it recommends webhooks where possible. If you're checking for new issues every few seconds so you can post them to a chat channel or a dashboard, a webhook will tell you the moment something happens and spend none of your allowance. This explainer on APIs versus webhooks covers the difference.
Want AI doing the heavy lifting in your marketing?
I build the systems that handle the boring 80 percent, so you get your week back. Done properly, with the human kept in.
How Datadog and PostHog publish limits
Datadog says many of its API endpoints are rate limited, a 429 response means you've been limited, and limits can be raised from the defaults by contacting support. PostHog's API docs list its analytics endpoints at 240 a minute and 1200 an hour, the events/values endpoint at 60 a minute and 300 an hour, and the query endpoint at 2400 an hour. If you're comparing monitoring stacks, my guide to API monitoring tools is a useful place to start.
My advice for integration builders
Don't hard-code a limit, because there isn't a single one to hard-code. Read the five headers on every response, track them per endpoint, and cap concurrency in your client. Replace any polling loop with a webhook, and don't bother creating extra tokens to get round a limit tied to who you are.
Want your tool on this list? Featured spots and reviews here are paid partnerships. Here's how to partner with me.
Frequently asked questions
What is the Sentry API rate limit?
It varies by endpoint. Sentry applies a fixed-window requests-per-second limit and a concurrent limit per caller and endpoint.
How do I see my remaining requests?
Read the X-Sentry-Rate-Limit-Remaining header on every response.
Can I get around it with more tokens?
No. The limiter uses the caller's identity, not the token.
How do I avoid hitting the limit?
Avoid polling and use Sentry webhooks where possible.
Is there a concurrent limit?
Yes, and the ConcurrentLimit and ConcurrentRemaining headers show it.

