Telegram channel scraper: posts, views, media and channel info from public channels

Telegram Channel Scraper pulls posts from any public Telegram channel through Telegram's own public web preview (https://t.me/s/<channel>). For each post you get the text, date, view count, media, links, forward and reply info. For each channel you get the title, description, subscriber count and photo. Results come back as clean JSON, CSV or Excel.

• No login, no phone number, no API keys. The Actor reads the same public page anyone can open in a browser. It doesn't use MTProto, user sessions or any Telegram account.
• Views as numbers. "5.83M" becomes 5830000, so you can sort and chart engagement straight away.
• History, not just the latest 20. The Actor pages backwards with ?before=<postId> until it reaches your maxPostsPerChannel, your since date or the first post of the channel.
• Monitoring mode. Turn on onlyNewSinceLastRun and schedule the Actor: each run returns (and charges for) only posts you haven't seen yet.
• $0.001 per post ($1 per 1,000). Private, missing or unreachable channels come back as free error rows.

Why use this Telegram channel scraper?

	Telegram Channel Scraper	Telegram API (MTProto) tools	Copy-paste by hand
Account needed	None	A phone number and API ID/hash; sessions can get banned	None
Setup	Paste channel names, click Start	Login flow, 2FA, session files	—
Views	Numeric views on every post	Yes	Visible, but "1.2K" text
History	Pages back as far as the channel's public preview goes	Yes	Slow scrolling
Personal data	None: channel-authored content only	Can expose member and commenter data	—
Price	$0.001 per post	Your servers plus account risk	Your time

Who uses it

• Media monitoring and PR teams: track what news, crypto or brand channels publish, and how many views each post gets.
• Researchers and analysts: build datasets of public channel posts by date range or keyword.
• Marketers: compare channels by subscribers and per-post views before you buy an ad placement.
• AI agents and RAG pipelines: feed fresh channel posts to an LLM with one tool call.

How to use it

1. Click Try for free. The form is prefilled with telegram and durov and 20 posts per channel.
2. Add the channels you want: bloomberg, @durov or https://t.me/s/telegram all work.
3. Optionally set Only posts since, Keywords or Only new posts since the last run.
4. Click Start. The prefilled run finishes in about 5 seconds. Open the Posts or Channels view, or export as JSON, CSV or Excel.
5. For monitoring, turn on Only new posts since the last run and add a Schedule (for example, hourly).

Input

Only channels is required.

Field	Type	What it does
channels	string[]	Public channel usernames, @names or t.me links. The same channel typed twice is scraped once. Invite links (t.me/+...), t.me/c/... links, users, bots and groups become free error rows.
maxPostsPerChannel	integer	Maximum posts per channel, newest first. Default 50.
since	string	Stop at posts older than this date. Accepts 2026-10-01, an ISO timestamp, or a relative period such as 7 days or 12 hours.
keywords	string[]	Keep only posts whose text, link-preview title or description, or document name contains these words. Case-insensitive.
keywordMode	any | all	any (default): at least one keyword must match. all: every keyword must match.
includeChannelInfo	boolean	Add one channel record per channel (title, description, subscribers, photo). Default true. Charged as channel-resolved. With onlyNewSinceLastRun, the record is added only when the channel has new posts.
onlyNewSinceLastRun	boolean	Skip posts this saved search returned before. History lives in a named key-value store.
stateKey	string	Optional monitor name. By default each keyword filter gets its own history.
maxPagesPerChannel	integer	Safety cap on preview pages (about 20 posts each) per channel. Default 50.
requestDelayMs	integer	Pause between page requests. Default 1000, minimum 500.
maxConcurrency	integer	Channels scraped in parallel. Default 2, maximum 5.

Input example: hourly monitor of three channels

{
  "channels": ["telegram", "durov", "https://t.me/s/bloomberg"],
  "maxPostsPerChannel": 100,
  "since": "7 days",
  "keywords": ["ai", "bitcoin", "update"],
  "onlyNewSinceLastRun": true,
  "stateKey": "tech-news-monitor"
}

Output

Every dataset item has a type: post, channel or error. Here is a real post from a local test run on 2026-10-22 (CDN URL and links shortened):

{
  "type": "post",
  "channel": "telegram",
  "channelTitle": "Telegram News",
  "postId": 460,
  "url": "https://t.me/telegram/460",
  "date": "2026-08-26T19:12:34.000Z",
  "text": "For all the info from today's update and a look back at 13 years of Telegram features — check out our blog:\nhttps://telegram.org/blog/welcome-messages-buttons-TG-13\n\nAugust Features\n1 • 2 • 3 • 4 • 5 • More",
  "views": 2020000,
  "edited": false,
  "hasMedia": false,
  "mediaTypes": [],
  "media": [],
  "linkPreview": {
    "url": "https://telegram.org/blog/welcome-messages-buttons-TG-13",
    "siteName": "Telegram",
    "title": "Welcome Messages, Buttons in Messages and Signed Gifts",
    "description": "Introducing Welcome Messages to greet new members of groups or channels, Buttons in Messages to make any text interactive, …",
    "imageUrl": "https://cdn4.telesco.pe/file/R_08vaIZo8g…"
  },
  "links": ["https://telegram.org/blog/welcome-messages-buttons-TG-13", "https://t.me/telegram/455", "…"],
  "isForwarded": false,
  "forwardedFrom": null,
  "forwardedFromUrl": null,
  "replyTo": null,
  "matchedKeywords": [],
  "isNew": null,
  "scrapedAt": "2026-10-22T05:09:58.580Z"
}

A channel record:

{
  "type": "channel",
  "channel": "durov",
  "url": "https://t.me/durov",
  "title": "Pavel Durov",
  "description": "Founder of Telegram.",
  "subscribers": 10300000,
  "photoUrl": "https://cdn4.telesco.pe/file/…jpg",
  "verified": true,
  "counters": { "photos": 102, "videos": 47, "files": null, "links": 200 },
  "scrapedAt": "2026-10-22T05:09:58.580Z"
}

An error row (free):

{ "type": "error", "channel": "no_such_channel_here", "input": "no_such_channel_here", "url": "https://t.me/no_such_channel_here", "errorCode": "CHANNEL_NOT_FOUND_OR_PRIVATE", "errorMessage": "…Nothing was charged." }

• Media. media[] lists each item with its type (photo, video, roundVideo, document, audio, voice, sticker, poll, location, unsupported), the image or video url when the preview exposes it, a video thumbnailUrl, and title/size for documents. Albums list every photo.
• Forwards. isForwarded is always set. forwardedFrom (a channel title) and forwardedFromUrl are filled only when the forward links to a public channel post. Names of individual users are never collected.
• Replies. replyTo holds the channel, postId and url of the post being answered, with no author names.
• Run summary. The run's OUTPUT record (key-value store) holds the status, each channel's health and stop reason, counts, billing and the onlyNew state.

The dataset has four ready-made views: Posts, Media & links, Channels and Skipped channels.

Pricing

Pay per event. Platform usage is included.

Event	Price	When it is charged
message-scraped	$0.001 per post	One post written to the dataset, after your since, keyword and onlyNew filters
channel-resolved	$0.003 per channel	One channel-info record written to the dataset (only when includeChannelInfo is on; in onlyNewSinceLastRun mode only when the channel has new posts)

Run	Records	Cost
Prefilled example (2 channels × 20 posts)	2 channels + 40 posts	$0.046
10 channels × 100 posts	10 channels + 1,000 posts	$1.03
Hourly monitor with onlyNewSinceLastRun, 5 channels	0–20 new posts (+ a channel record per channel with news)	$0 when nothing is new; ~$0.003–$0.04 otherwise

You are never charged for:

• private, missing, invalid or unreachable channels (error rows);
• posts removed by your since, keyword or onlyNew filters;
• service messages ("channel created", "photo updated").

Set Max cost per run (maxTotalChargeUsd) and the Actor stops cleanly at that budget: it pushes only records it can charge for and marks the run budget_limit_reached. If the budget can't cover even the first record ($0.003 for a channel record, or $0.001 for a post when includeChannelInfo is off), the run ends at once with BUDGET_TOO_LOW and fetches nothing.

API and integrations

Copy-paste ready: set APIFY_TOKEN to your own token from Apify Console → Settings → API & Integrations.

HTTP (one call, returns the records):

curl -X POST "https://api.apify.com/v2/acts/impressionable_lupine~telegram-channel-scraper/run-sync-get-dataset-items?token=<APIFY_TOKEN>" \
  -H "Content-Type: application/json" \
  -d '{"channels":["telegram","durov"],"maxPostsPerChannel":20}'

JavaScript:

import { ApifyClient } from 'apify-client';
const client = new ApifyClient({ token: process.env.APIFY_TOKEN });
const run = await client.actor('impressionable_lupine/telegram-channel-scraper').call({ channels: ['bloomberg'], since: '3 days', includeChannelInfo: false });
const { items } = await client.dataset(run.defaultDatasetId).listItems();

Python:

from apify_client import ApifyClient
client = ApifyClient("<APIFY_TOKEN>")
run = client.actor("impressionable_lupine/telegram-channel-scraper").call(run_input={"channels": ["durov"], "keywords": ["ton"], "maxPostsPerChannel": 200})
posts = [i for i in client.dataset(run["defaultDatasetId"]).list_items().items if i["type"] == "post"]

Results also flow to Zapier, Make, n8n, Google Sheets, Slack and webhooks through Apify integrations. A common setup is an hourly schedule with onlyNewSinceLastRun, plus a webhook that posts the new items to Slack.

Use with AI agents (MCP)

The Actor works as a tool through the Apify MCP server in Claude, Cursor, VS Code and any MCP client:

{
  "mcpServers": {
    "apify": {
      "url": "https://mcp.apify.com/?tools=impressionable_lupine/telegram-channel-scraper",
      "headers": { "Authorization": "Bearer <APIFY_TOKEN>" }
    }
  }
}

Example prompt: "Use telegram-channel-scraper to get the last 30 posts from @durov and @telegram, then list the five most-viewed posts with their date, views and link."

Agents get a flat, predictable schema with numeric views, ISO dates and canonical urls, at $0.001 per post.

Limitations

• Public channels only. The Actor reads t.me/s/<channel>, which exists only for public channels whose owners haven't turned off the web preview. Private channels, invite links, groups, users and bots are skipped as free error rows.
• What the preview shows. View counts are rounded as Telegram displays them ("1.2K" → 1200). Reactions, comments and comment counts aren't collected. Some media (rich posts, some animations) shows as unsupported with a link to the post.
• Media URLs expire. Photo and video URLs point to Telegram's CDN and are signed. Download what you need soon after the run.
• History depth. Paging goes back as far as the public preview allows. Very large backfills take time, because the Actor waits requestDelayMs between pages.
• Forward sources. The preview often hides the source of a forward. forwardedFrom is filled only when it links to a public channel.

FAQ

Is it legal to scrape Telegram channels? This Actor reads only public channel pages that Telegram serves to anyone without login, and t.me publishes no robots.txt restrictions (checked 2026-10-22). It collects content the channel itself published, not personal data about subscribers or commenters. You're still responsible for how you use the data: respect copyright, Telegram's Terms of Service and privacy laws such as GDPR, and don't use the output to profile individuals. This isn't legal advice; if in doubt, ask a lawyer.

Can it scrape private channels or groups? No. It has no login and doesn't join chats. Private channels, invite links (t.me/+...), t.me/c/... links and groups are reported as error rows and never charged.

Does it collect user data or phone numbers? No. There are no member lists, commenter profiles or phone numbers. Post signatures (admin names), the names of users that posts were forwarded from, and reply author names are deliberately left out.

Why is a channel showing CHANNEL_NOT_FOUND_OR_PRIVATE when it opens in my app? The channel may be private, restricted in your region, have the web preview turned off, or the username may belong to a user, bot or group. Check that https://t.me/s/<name> opens in a browser.

How does "only new posts" work? For each channel, the Actor stores the IDs of the posts it returned (and the highest ID) in a named key-value store (telegram-channel-scraper-state), under a key derived from your keyword filter or your stateKey. The next run stops paging when it reaches posts it has already seen, so you pay only for new ones.

How fast is it? About 1 second per page of 20 posts per channel, with 2 channels in parallel by default. 1,000 posts from one channel take about a minute.

Do I need proxies? No. The Actor makes a few polite requests per channel and uses no proxies.