X (Twitter) Replies Scraper - Tweet Replies, No Login avatar

X (Twitter) Replies Scraper - Tweet Replies, No Login

Pricing

from $3.50 / 1,000 x/twitter reply scrapeds

Go to Apify Store
X (Twitter) Replies Scraper - Tweet Replies, No Login

X (Twitter) Replies Scraper - Tweet Replies, No Login

Export public replies to X (Twitter) posts from tweet URLs or IDs. Each reply row has text, author profile (followers, bio, verified), likes, views, bookmarks, language and conversation IDs. Sort by relevance, recency or likes and resume big exports with a saved cursor. No login or cookies.

Pricing

from $3.50 / 1,000 x/twitter reply scrapeds

Rating

0.0

(0)

Developer

Khadin Akbar

Khadin Akbar

Maintained by Community

Actor stats

0

Bookmarked

25

Total users

3

Monthly active users

7 days ago

Last modified

Features

Limited permissions

Share

What is X (Twitter) Replies Scraper?

X (Twitter) Replies Scraper exports the public replies to any X (Twitter) post, with no login or cookies. Give it tweet URLs or numeric tweet IDs (up to 50 per run) and get one dataset row per reply: reply text, time, language, the replying account (followers, following, bio, location, verified and blue-check status, account age), and likes, views, retweets, quotes and bookmarks.

Sort replies by Relevance, Recency or Likes. Keep only replies that pass your filters (likes, views, replies, retweets, language, date range, verified accounts, skip the post author) before you are charged, read a second level of replies-to-replies, and resume large exports with a cursor. Every run also writes a short agent summary (top accounts, language mix, top replies) so an AI agent can answer without reading every row.

What you getFields
ReplyreplyId, replyUrl, text, createdAt, language, conversationId, inReplyToTweetId, source, depth, parentReplyId
Replying accountauthorUsername, authorName, authorId, authorUrl, authorVerified, authorBlueVerified, authorFollowers, authorFollowing, authorBio, authorLocation, authorAccountCreatedAt, authorProfileImageUrl
EngagementlikeCount, replyCount, retweetCount, quoteCount, viewCount, bookmarkCount
Contenthashtags, mentions, urls, mediaUrls
ContextparentTweetId, parentTweetUrl, rankingMode, position, page, scrapedAt
Post summary (free, in OUTPUT)text, author, counts and views of each input post

What's new in 1.1

  • Faster: replies are read 36 per page (was about 20) and up to 4 posts are read in parallel.
  • Depth 2: set maxDepth: 2 to also get replies to replies (up to 25 reply threads per post).
  • More filters: minimum views, replies and retweets, date range (since, until), verified accounts only, skip the post author's own replies.
  • Richer rows: author location, blue-check status, account creation date, reply depth.
  • Safer costs: a scan limit stops a run that reads many replies but keeps few, so strict filters cannot drain a run (see Pricing).
  • Free post summary and agent summary in OUTPUT.
  • Lower prices with Store discounts: the price per reply now drops at every Apify plan level.

Best fit

  • Launch and campaign reviews: read what people replied to an announcement.
  • Brand monitoring: pull the newest replies to your posts every day with rankingMode: "Recency" and since.
  • Social listening and sentiment: export reply text with language and author reach for analysis.
  • Lead research: find engaged accounts with large followings; use minViews, verifiedOnly and authorFollowers.
  • AI agents: one call returns flat reply rows plus an agentSummary ready for summarising.

Which X (Twitter) Actor should I use?

This Actor is part of a family of public X (Twitter) Actors that share the same no-login approach and similar row fields. Pick the one that matches your job:

Practical scenario

A community manager wants the most-liked replies to a product announcement before the weekly review. She enters the post URL, sets rankingMode: "Likes", minLikes: 5, excludeOriginalAuthor: true and maxReplies: 200. Low-signal replies and the brand's own answers are skipped and not charged. She sorts the export by likeCount, reads the top themes in text, and checks authorFollowers to spot influential voices.

Input

FieldTypeDefaultDescription
tweetUrlsarrayrequiredTweet URLs (https://x.com/OpenAI/status/2065225362544726371) or numeric tweet IDs, up to 50.
rankingModestringRelevanceRelevance, Recency (newest first) or Likes (most liked first).
maxRepliesinteger100Total reply rows saved and charged across all posts (1–5000).
maxDepthinteger11 = replies to the post. 2 = also replies to replies (up to 25 threads per post).
minLikesinteger0Keep only replies with at least this many likes.
minViewsinteger0Keep only replies seen at least this many times.
minRepliesinteger0Keep only replies that got at least this many replies.
minRetweetsinteger0Keep only replies retweeted at least this many times.
languagestringemptyKeep only replies in one language, as a two-letter code such as en.
since / untilstringemptyKeep replies posted on or after / on or before a date (YYYY-MM-DD, UTC).
verifiedOnlybooleanfalseKeep only replies from verified or blue-check accounts.
excludeOriginalAuthorbooleanfalseSkip replies written by the post's own author.
maxPagesPerTweetinteger10How deep to follow each conversation (1–100 pages of up to 36 replies).
startCursorstringemptyContinue the first post from RUN_SUMMARY.nextCursors.
dedupeRepliesbooleantrueSkip duplicate replies across pages and posts.

Rows skipped by a filter or by dedupe are never charged.

{
"tweetUrls": ["https://x.com/OpenAI/status/2065225362544726371"],
"rankingMode": "Likes",
"maxReplies": 200,
"minLikes": 5,
"language": "en",
"excludeOriginalAuthor": true
}

Output

One dataset row per public reply. Sample row (replying account anonymised; other values from a real export):

{
"parentTweetId": "2065225362544726371",
"parentTweetUrl": "https://x.com/OpenAI/status/2065225362544726371",
"rankingMode": "Relevance",
"position": 1,
"depth": 1,
"replyId": "2099602156332429431",
"replyUrl": "https://x.com/example_user/status/2099602156332429431",
"text": "@OpenAI Please also enable the ability to buy a usage reset.",
"createdAt": "2026-09-14T20:52:17.000Z",
"language": "en",
"conversationId": "2065225362544726371",
"inReplyToTweetId": "2065225362544726371",
"authorUsername": "example_user",
"authorName": "Example User",
"authorUrl": "https://x.com/example_user",
"authorVerified": false,
"authorBlueVerified": true,
"authorFollowers": 80,
"authorFollowing": 26,
"authorLocation": "Berlin",
"authorAccountCreatedAt": "2019-03-02T10:11:12.000Z",
"replyCount": 0,
"retweetCount": 0,
"quoteCount": 0,
"likeCount": 2,
"viewCount": 34,
"bookmarkCount": 0,
"mentions": ["openai"],
"page": 1,
"scrapedAt": "2026-11-15T06:53:21.824Z"
}

Each run also writes OUTPUT and RUN_SUMMARY to the default key-value store with an outcome (COMPLETE, PARTIAL, VALID_EMPTY, INVALID_INPUT or UPSTREAM_FAILED), counts, replies skipped by filters, postSummaries (one free summary per input post), agentSummary (top accounts by followers, language mix, top replies, next step) and nextCursors for resuming.

Pricing

Pay per event, with Apify platform usage included. You pay the x-reply-scraped event once for each reply saved to the dataset, plus the small apify-actor-start event. The price per reply drops as your Apify plan level rises:

Apify planPrice per reply1,000 replies
Free$0.0016$1.60
Bronze$0.0014$1.40
Silver$0.0012$1.20
Gold and above$0.0010$1.00

Examples at the Free price: 100 replies cost about $0.16; 1,000 replies about $1.60. Replies skipped by filters or dedupe are never charged, and maxReplies is a hard cap on reply events. Replies are read 36 at a time, so a run can read a few rows more than it saves; only saved rows are charged. Scan limit: if your filters are so strict that the Actor reads far more replies than it keeps (more than 1.5 reads per saved row after the first 72), it stops with stopReason: "filters-too-selective" and outcome: "PARTIAL" (or VALID_EMPTY if nothing matched), so a run can never silently chew through a large conversation. Loosen the filters to read further. See the Pricing tab for the current event prices.

Use with AI agents (MCP)

Actor identity: khadinakbar/x-twitter-replies-scraper.

Get the 100 most-liked English replies to https://x.com/OpenAI/status/2065225362544726371 and summarise the main requests people make.

{ "tweetUrls": ["https://x.com/OpenAI/status/2065225362544726371"], "rankingMode": "Likes", "maxReplies": 100, "language": "en" }

Read OUTPUT.outcome first. VALID_EMPTY means the post has no matching public replies; PARTIAL means some rows were saved and RUN_SUMMARY.nextCursors or the scan limit explains why it stopped; INVALID_INPUT lists what to fix in invalidInputs. OUTPUT.agentSummary gives top accounts, language mix and top replies without reading the dataset. replyUrl links each row back to X.

API example

import { ApifyClient } from 'apify-client';
const client = new ApifyClient({ token: process.env.APIFY_TOKEN });
const run = await client.actor('khadinakbar/x-twitter-replies-scraper').call({
tweetUrls: ['https://x.com/OpenAI/status/2065225362544726371'],
rankingMode: 'Recency',
maxReplies: 50,
since: '2026-06-12',
});
const { items } = await client.dataset(run.defaultDatasetId).listItems({ clean: true });
console.log(items.map((r) => [r.authorUsername, r.likeCount, r.text]));

Best results

  • Use Likes for the strongest replies, Recency for monitoring and Relevance for X's default order.
  • Likes order with minLikes, and Recency order with since, stop early once later replies cannot match.
  • Start with maxReplies: 50 to check a conversation, then raise it.
  • Use maxDepth: 2 only when you need sub-threads; each sub-thread read costs a little extra time.
  • For very large threads, save RUN_SUMMARY.nextCursors and pass it as startCursor in the next run with the same post and ranking mode.

Builder's note

I built this for workflows that start from one known post. I found that most buyers want a few hundred strong replies, not every one-word reaction, so every filter runs before billing: you pay only for rows that pass. Version 1.1 reads more replies per page and runs posts in parallel so large exports finish faster, and lowers the price at every plan level.

FAQ

Do I need an X account or cookies?

No. The Actor reads public reply data only.

Why did I get fewer replies than the post shows?

X hides some replies (for example from limited or deleted accounts, and some low-quality or hidden replies) from public view. The Actor saves the public replies it can read; raise maxPagesPerTweet to follow the conversation deeper. Hidden replies cannot be returned.

Can I continue a big export?

Yes. Copy the cursor from RUN_SUMMARY.nextCursors into startCursor and run again with the same post and ranking mode. A cursor works for the first post in your list.

Does it return nested replies?

Yes, one extra level. Set maxDepth: 2; rows then carry depth: 2 and parentReplyId. Deeper levels are not collected. Depth-2 reads look at up to 25 reply threads per post, those with the most replies first.

What happens if my filters are too strict?

The run stops with stopReason: "filters-too-selective" instead of reading the whole conversation. Rows already saved stay charged once; unmatched rows are never charged.

Responsible use

This Actor collects publicly available X (Twitter) replies. Use the data in line with X's terms, privacy law and your own data policies, especially when handling personal profile data. This Actor is not affiliated with or endorsed by X Corp.