Telegram Channel Scraper - posts, views, reactions and exact subscriber counts from public channels, no login

Pull the posts of any public Telegram channel and watch them over time: full text, date, views, reactions (including paid Stars), photos, video thumbnails, files, link previews, links, hashtags, forwards and replies, plus the channel's exact subscriber count. You can search inside a channel, fetch only the posts published on or after a date, or switch to a channels mode with one row per channel.

It reads the same public page anyone sees in a browser at t.me/s/<channel>. No Telegram account, no phone number, no API key, no login, no cookies.

Read this first

• Public channels only. Users, bots, groups and private channels (invite links such as t.me/+AbC...) have no public preview. They are reported in FAILURES (no_public_preview or unsupported_input) and are never scraped and never charged.
• No member lists, no comments, no phone numbers. This actor does not read who subscribed, who commented or who replied. A reply only keeps the link to the post it answers (replyToUrl).
• E-mails and phone numbers in common formats are replaced (best effort) with [email redacted] and [phone redacted] in post text, channel description, link-preview text, file names and link addresses. Recognised: e-mail addresses (also spelled-out forms such as name (at) host.com), phone numbers that start with + or 00, UK, German, French and US/Canada formats, wa.me/<number> and tel: links, and Russian and other CIS numbers such as +7 495 123-45-67, 8 (495) 123-45-67, 8 800 555 35 35 and 89123456789. Not recognised, so they stay as published: other countries' national formats written without a + (for example India, Brazil, China, Singapore), bare ten- or eleven-digit numbers written without separators (apart from eleven digits that start with 8, such as 89123456789) and e-mail addresses with non-Latin letters (ivan@почта.рф). Channel titles and author signatures are checked for e-mail addresses only, and link-preview site names are not checked. The actor is not a contact-harvesting tool: it never reads user pages or member lists.
• Views and the preview's subscriber count are rounded by Telegram above 1,000 (it shows 5.82M). views is therefore approximate above 1,000; below that it is exact. subscribers is the exact number from the channel's landing page; it is null if that page did not answer or you switched exactSubscribers off, and subscribersApprox (the rounded preview number) is always there.
• History goes as far back as the channel's web preview allows. A page holds up to 20 posts, newest first (fewer where messages in between were deleted, hidden or are stickers). The actor follows the preview's own "older posts" link until the channel's first post, your post limit, or your date limit.
• Posts without text, photo, video, file or link card are skipped and not charged. That covers service messages (channel photo changed, pinned message, "channel created"), posts that are only a sticker, and other posts the web preview cannot show as text or a file (the preview then says "open Telegram to view this post"). They are never delivered and never charged, and the run summary counts them: skippedService (service messages), skippedStickerOnly (sticker-only posts) and skippedEmpty (everything else). A channel that posts only stickers therefore gives 0 rows and a skippedStickerOnly count.
• A refusal from Telegram ends the run. If Telegram answers 429, 403 or any other refusal, the actor stops, keeps every row already collected and says blocked in the run summary. It does not retry, rotate or disguise itself.

What you get

Posts mode (default): one row per post

field	meaning
postId, postUrl	the post's number and its https://t.me/<channel>/<id> link
channelUsername	the channel as Telegram spells it (the capitalisation can differ from your input)
date	publication time, ISO with offset (2026-10-11T14:52:40+00:00)
edited	true if the post shows the "edited" mark
text	full text, line breaks kept, e-mails and phones in common formats replaced (best effort)
viewsText, views	views as shown (1.2M) and as a number (1200000); approximate above 1,000
reactions, reactionsTotal	each reaction {emoji, customEmojiId, paidStars, count} and their sum. Empty list and 0 when the post shows none (also when the channel switched reactions off)
mediaType	album, video, photo, document, link or text
photoUrls, videoCount, videoThumbnailUrls	photo addresses, number of videos, video thumbnail addresses (Telegram CDN links; they can change over time, so keep the files if you need them)
documents	attached files as {title, extra} (file name and size)
linkPreview	{url, siteName, title, description} of the link card, or null
links, hashtags	web links and hashtags written in the text
authorSignature	the author's signature if the channel adds one
isForwarded	true for forwarded posts
forwardedFrom	{name, url}: the original source and the link to the original post, as the preview shows them for a forward from a public channel. null when the preview shows no source (forwards out of users and hidden sources) or the post is not a forward
replyToUrl	link to the post this one replies to
channelTitle, channelVerified	the channel's name and verified badge
subscribers, subscribersApprox	exact subscriber count (10315764) and the preview's rounded count (10300000)
searchQuery, input, scrapedAt	the search you used (or null), the channel entry as you typed it, and the time of the run

Channels mode: one row per channel

Set outputMode to channels. Two requests per channel: channelUsername, channelUrl, channelTitle, channelDescription, channelVerified, subscribers (exact), subscribersText, subscribersApprox, photosCount, videosCount, filesCount, linksCount, latestPostId, latestPostDate, postsSampled, avgViewsLatest (average views of the latest posts on the first page), channelPhotoUrl (a web address, or null for a channel without a photo), input, scrapedAt. Good for ranking or monitoring many channels cheaply.

Input

field	default	meaning
channels (required)	-	usernames or links: durov, @durov, https://t.me/durov, https://t.me/s/durov. Duplicates (any capitalisation) are read once
outputMode	posts	posts or channels
searchQuery	empty	Telegram's own search inside each channel's public preview (posts mode only)
maxPostsPerChannel	100	newest first, hard cap 5000 (posts mode only)
postedAfter	empty	YYYY-MM-DD (UTC): only posts on or after that day; paging stops at the first older post (posts mode only). The date is fixed in the saved input, so a schedule repeats it: every scheduled run delivers everything since that date again. For new posts only, set the date from your own scheduler or API call (for example yesterday's date), or drop repeats by postUrl (or channelUsername + postId) on your side
exactSubscribers	on	one extra request per channel for the exact subscriber count
maxResults	1000	rows per run across all channels, hard cap 100000
proxyConfiguration	Apify datacenter proxy	the default is what this actor was tested with

Example: {"channels": ["durov", "https://t.me/telegram"], "maxPostsPerChannel": 100} returns up to 200 posts. {"channels": ["durov"], "searchQuery": "ton", "maxPostsPerChannel": 30} returns the channel's posts that match "ton".

Example row

{
  "postId": 552,
  "postUrl": "https://t.me/durov/552",
  "channelUsername": "durov",
  "date": "2026-10-11T14:52:40+00:00",
  "edited": false,
  "text": "✨ Designers × AI\n\n🆕 The mockup era is over. ...",
  "viewsText": "1.2M",
  "views": 1200000,
  "reactionsTotal": 29120,
  "reactions": [{"emoji": null, "customEmojiId": null, "paidStars": true, "count": 5950}, {"emoji": null, "customEmojiId": "5936157098181135162", "paidStars": false, "count": 13200}],
  "mediaType": "video",
  "photoUrls": [],
  "videoCount": 1,
  "videoThumbnailUrls": ["https://cdn4.telesco.pe/file/VnHVmW499ucIS1pVBdP5pJ5ag5FLYR8N..."],
  "documents": [],
  "linkPreview": null,
  "links": ["https://t.me/designers"],
  "hashtags": [],
  "authorSignature": "Pavel Durov",
  "isForwarded": true,
  "forwardedFrom": null,
  "replyToUrl": null,
  "channelTitle": "Pavel Durov",
  "channelVerified": true,
  "subscribers": 10315764,
  "subscribersApprox": 10300000,
  "searchQuery": null,
  "input": "durov",
  "scrapedAt": "2026-10-20T10:00:00.000Z"
}

(The text, reactions and thumbnail address are shortened here; real rows carry the full values.)

Speed, limits and how a run ends

• One request every 2 seconds, one after the other, plus the time to fetch and read each page. A page holds up to 20 posts. Measured on the platform on 21 Oct 2026 at 512 MB of memory on two big news channels: about 3 seconds per page (runs of 32 and 52 requests: 2.9 and 3.0 seconds per request, the 2-second pause included), so 100 posts take about 15 to 20 seconds when every page holds 20 posts; a third, small run took 8.7 seconds per page when Telegram answered slowly. Channels with deleted, hidden or sticker-only messages have fewer posts per page (7 pages for 28 posts and 10 pages for 40 posts in two small channels), so allow more requests there. The figure for 5,000 posts (about 250 pages, 13 to 20 minutes per channel) is an estimate, not a measured run. One more request per channel asks for the exact subscriber count. Less memory is slower, because reading the pages needs CPU.
• A very large run (many channels with a high maxPostsPerChannel) can take hours. Raise the run timeout for those. The actor stops starting new requests shortly before the timeout, keeps every row already collected and reports time_budget in the run summary, so you can see that the result is partial.
• RUN_SUMMARY (key-value store record) is written at the end of every run that reads Telegram: {complete, terminationReason, rowsPushed, channels, noPreview, skippedService, skippedEmpty, skippedStickerOnly, requests}. A run that stops at once because no input was usable writes only FAILURES. If the platform restarted the run, restarts is added as the last key.
  • terminationReason: finished (everything asked for was read), max_results (your maxResults was reached), charge_limit (your maximum cost per run was reached), time_budget (the run timeout was near), blocked (Telegram refused a request), request_failed (a page could not be read after two retries, so the result is partial). complete is true only for finished and max_results.
  • channels counts channels whose preview was read, noPreview counts inputs without a public preview.
• FAILURES (key-value store record) lists the inputs that could not be read: {input, channel, status, httpStatus} with status one of unsupported_input (not a public channel username or t.me link, including private invite links), no_public_preview (a user, bot, group, unknown name or a channel without a public preview), not_found (HTTP 404), request_failed (server error or timeout after two retries), blocked (Telegram refused the request). A channel that was read but has nothing to deliver (only stickers, no search match, nothing newer than postedAfter) is not listed: check RUN_SUMMARY (rowsPushed, skippedStickerOnly). A channel that gave rows can also appear here when a later page failed.

Pricing

Pay per result, at the rate shown in the pricing panel of this page, plus a very small per-run start fee. You are charged only for the results delivered.

Only delivered rows are charged: channels without a public preview, service messages, empty messages and failed requests cost nothing.

Good to know

• The actor reads Telegram's public web preview. If Telegram changes that page, fields can become empty before this actor is updated; the run summary and FAILURES show it.
• Channels can hide older posts, delete posts or switch reactions off; you get what the preview shows on the day of the run.
• If Apify moves the run to another server, the run restarts: it reads its pages again but never writes (or charges) a post it already delivered. A restart waits about 12 seconds before it continues, and the rows still come out once.
• Use it for channel monitoring, crypto and news channel tracking, competitor and brand mentions, content research and subscriber-count tracking over time. For alerts, schedule a run with searchQuery and set postedAfter yourself (a schedule repeats a fixed date), or drop repeats by postUrl.

Keywords: telegram scraper, telegram channel scraper, telegram posts scraper, telegram messages, telegram channel monitoring, telegram subscribers count, t.me scraper, crypto telegram channels, telegram analytics, no telegram api.