Skip to content

Twitter API rate limiting: limits, errors and fixes

Twitter API rate limiting explained: X's 15-minute windows, per-app and per-user limits, the 429 error, the headers to read and how to stop hitting the limit.

Piyushh · November 13, 2026 · 3 min read

A canal lock holding back a queue of paper boats, picturing API rate limiting.

Twitter API rate limiting means X caps how many requests you can make in a time window, usually 15 minutes. Go over and the API returns HTTP 429, and you wait for the window to reset. There is no single number, because the limit depends on the endpoint and on how you authenticate. Below: how the limits are built, where to read your own, and how to stop hitting them.

Everything about X below comes from X's rate limits page, checked 2026-11-13. X can change these numbers, so confirm them there before you rely on them.

How the Twitter API rate works

Most of it comes down to three ideas.

Limits count requests in a window, usually 15 minutes. X's page says some endpoints use 24 hours or per second instead.

Limits are set per app and per user, and the page treats the two cases separately. Per-app limits apply when you authenticate with a Bearer Token (app-only). Per-user limits apply with OAuth 1.0a or OAuth 2.0 user tokens. Each endpoint has its own number for each case. If you are not sure which one you are using, read our guide on Twitter API authentication.

Each endpoint keeps its own counter, so using up search does not use up tweet lookup.

Twitter API request limits for common endpoints

X's page gives these examples, per 15 minutes:

EndpointPer appPer user
Recent search, /2/tweets/search/recent450300
Tweet lookup, /2/tweets3,5005,000
User timeline, /2/users/:id/tweets10,000900

Other endpoints have their own docs pages. Check the page for the endpoint you call rather than a blog post, this one included.

The 429 error and the headers

X's page says that when you go over a limit you get a 429 with error code 88 and the message "Rate limit exceeded". Three response headers tell you where you stand:

  • x-rate-limit-limit: the maximum requests allowed
  • x-rate-limit-remaining: requests left in the window
  • x-rate-limit-reset: a Unix timestamp for when the window resets

Read them on every response, not only on failures, so you slow down before the 429 instead of after it.

If you're dealing with the error itself, our post on the X rate limit exceeded error covers the causes. For other X access errors such as 403, see Twitter API restrictions.

A rate limit is not a spending limit

A rate limit and a spending limit are easy to confuse. A rate limit means "too many requests, wait". A spending limit means "no more budget". On X's pay-per-use API, running out of credits is its own failure with its own error, covered in is the X API free. When your script stops, work out which limit you hit before you change any code.

How to stop hitting the limit

  1. Read the headers and sleep. When x-rate-limit-remaining reaches 0, sleep until x-rate-limit-reset.
  2. Back off on a 429. Wait, retry, and double the wait each time, up to a cap.
  3. Ask for more in each request. Use the largest page size the endpoint allows, so you make fewer calls.
  4. Cache results. If two parts of your program ask the same question, answer the second one from memory.
  5. Spread the work out. A queue running at a steady rate does better than a burst followed by a long wait.
  6. Stop polling things that haven't changed. Check a quiet account less often than a busy one.

Here is a small retry helper in Python:

import time, requests

def get_with_backoff(url, headers, params=None, tries=5):
    wait = 2
    for _ in range(tries):
        r = requests.get(url, headers=headers, params=params, timeout=60)
        if r.status_code != 429:
            r.raise_for_status()
            return r.json()
        reset = r.headers.get("x-rate-limit-reset")
        if reset:
            wait = max(wait, int(reset) - int(time.time()) + 1)
        time.sleep(min(wait, 900))
        wait *= 2
    raise RuntimeError("still rate limited after retries")

It uses X's own reset header when the response includes one, and doubles the wait when it doesn't.

Rate limits on a hosted API

On a hosted API instead of X's own, the limits are set by that API. ChatterAPI is independent and not affiliated with X Corp. It limits requests per second by plan: 10 on Starter, 25 on Growth, 50 on Scale. Failed calls cost no credits. A search call looks like this:

curl -s "https://api.chatterapi.com/v1/x/tweets/search?q=rate%20limit&sort=latest" \
  -H "Authorization: Bearer $CHATTERAPI_KEY"

Every response carries a credits object showing what the call cost and what is left, so you can track your budget the way you would track the X headers. The X search endpoint docs list the parameters.

Rate limits still exist. They change from 15-minute windows set by X to per-second limits set by your plan, and you no longer need a separate X developer app or a separate bill.

Check the limits you are on

Our own limits and plan prices are on the pricing page.

FAQ

What is the Twitter API rate limit?
It depends on the endpoint and on how you authenticate. X's docs, checked 2026-11-13, list for example 450 requests per 15 minutes per app for recent search, and 300 per user. Each endpoint has its own numbers on its docs page.
What does a Twitter API 429 error mean?
You sent more requests than your window allows. X returns status 429 with the message Rate limit exceeded. Wait until the reset time in the response headers, then retry.
How do I see how many requests I have left?
Read the three rate limit headers on every X response: x-rate-limit-limit, x-rate-limit-remaining and x-rate-limit-reset. The last one is a Unix timestamp for when the window resets.
Is the Twitter API request limit the same as the spending limit?
No. A rate limit caps how many requests you can send in a time window. A spending limit caps how much you pay. You can hit either one, and the fixes differ.
Does ChatterAPI have rate limits?
Yes, by plan, measured in requests per second. They are listed on the pricing page. Failed calls cost no credits.

More on X (Twitter) API

Try it free for 7 days.

1,000 credits to make your first calls. One key for Reddit and X, over REST or MCP.