Twitter Scraper — Search Tweets by Keyword

Search Twitter (X) by keyword, hashtag or advanced operator and export every matching tweet — text, timestamp, language, likes, reposts, replies, quotes, bookmarks, views, author and attached media — as one row per tweet.

Give it one query or fifty. Each runs in turn, newest first or most engaged first, until you hit your cap or the results run out.

Input

Field	Type	Default	What it does
searchQueries	list of text	—	Required. One or more searches. Operators go inside the text.
sort	latest / top	latest	latest is newest first. top is the most engaged first.
maxItems	integer	100	Caps the whole run, across every query. 0 means no limit.

{
  "searchQueries": ["from:openai", "openai lang:en"],
  "sort": "top",
  "maxItems": 100
}

Operators you can put in a query:

Operator	Matches
from:handle	Tweets written by that account
"exact phrase"	The phrase, word for word
#hashtag	Tweets carrying the hashtag
lang:en	Tweets X detected as English
min_faves:100	Tweets with at least 100 likes
since:2026-10-01 until:2026-10-15	A date window
-filter:replies	Everything except replies
(a OR b)	Either term. Every ( needs its ).

Output

One row per tweet.

Field	What it holds
id, url	The tweet's ID and its link on x.com
text, lang	What it says, and the language X detected
created_at	Publish time, ISO 8601 UTC
like_count, reply_count, repost_count, quote_count, bookmark_count, view_count	The six counters, as they read when the tweet was fetched
author_handle, author_name, author_id, author_followers_count, author_is_blue_verified	The author, flattened for spreadsheets
author	The author's full profile record
media, hashtags, mentions, urls, card, article, place	Photos, videos and GIFs, hashtags, mentioned accounts, links with their destinations, link card, attached article, tagged place
is_reply, in_reply_to_post_id, in_reply_to_handle, conversation_id	Where the tweet sits in a thread
quoted_post, reposted_post	The tweet it quotes, or the original it reposts
input_query	The query that found it, so several searches can share one dataset

{
  "id": "2110399768232898915",
  "url": "https://x.com/zerohedge/status/2110399768232898915",
  "text": "*OPENAI COMPUTING CHIEF TO DEPART: WSJ",
  "lang": "en",
  "source": "Twitter for iPhone",
  "created_at": "2026-10-14T15:58:08.000Z",
  "like_count": 881,
  "reply_count": 39,
  "repost_count": 57,
  "quote_count": 28,
  "view_count": 175617,
  "bookmark_count": 52,
  "is_reply": false,
  "in_reply_to_post_id": null,
  "in_reply_to_user_id": null,
  "in_reply_to_handle": null,
  "conversation_id": "2110399768232898915",
  "is_pinned": null,
  "hashtags": [],
  "author_id": "18856867",
  "author_handle": "zerohedge",
  "author_name": "zerohedge",
  "author_followers_count": 3432237,
  "author_is_blue_verified": true,
  "input_query": "openai lang:en"
}

The author, media, quoted_post, reposted_post and the other nested fields are cut from this example to keep it short. They are in every row.

Pricing

You pay for each tweet saved. A tweet that is skipped is never charged.

Plan	Per 1,000 tweets
Free	$2.00
Bronze	$1.69
Silver	$1.38
Gold and above	$1.08

Each run also carries a $0.00005 start charge. The Free plan covers the first 50 rows you save with this Actor; after that you need a paid plan.

Questions

Why did I get fewer tweets than maxItems? A query stops when X stops returning pages for it or returns an empty one. Broad queries get cut off well before they run dry. Split them into date windows — since: and until: — and give each its own entry in searchQueries.

Does maxItems apply per query or to the whole run? The whole run, in the order you list the queries. A busy first query can use the entire budget. Put the one you care about first, or run the queries separately.

What do latest and top change? latest walks back from the newest tweet and goes as far as X lets it. top ranks by engagement and usually ends sooner. Use latest for monitoring and top for "what did people react to".

A row's text starts with "RT @". Where is the full tweet? In reposted_post. On a repost, text is X's shortened form. A quote tweet keeps its own text, and the tweet it quotes sits in quoted_post.

What happens to a query X rejects? It is skipped, and the log says why — an unbalanced parenthesis is one cause. Other queries in the list still run. If every query is rejected, the run fails with that reason and nothing is charged.

Can I run this from code or on a schedule? Yes. Any Apify client works, and Apify's scheduler can run it hourly or daily:

from apify_client import ApifyClient

client = ApifyClient("<YOUR_APIFY_TOKEN>")
run = client.actor("toolzerhub/twitter-scraper").call(
    run_input={"searchQueries": ["from:openai"], "maxItems": 50}
)
for row in client.dataset(run["defaultDatasetId"]).iterate_items():
    print(row["created_at"], row["text"])

Are the counters exact? They are live numbers. Two runs a few minutes apart will disagree on likes and views. That is drift, not a fault.

Related Actors

Actor	Use it when
Twitter Tweets Scraper	You want one account's own timeline rather than a search
Twitter Comments Scraper	You already have a tweet and want the replies under it
Twitter Profile Scraper	You need the full profile behind an author_handle
Twitter Followers Scraper	You want the accounts that follow an author

Support

Questions, bugs, or feature requests: contact@toolzerhub.com

Browse the rest: apify.com/toolzerhub