Telegram Channel Scraper & Keyword Alerts | $0.50/1K

What does Telegram Channel Scraper & Keyword Alerts do?

It reads posts from public Telegram channels and returns them as clean, structured data, with no Telegram account, no phone number and no API keys. It can also watch channels for you: every run returns only the posts published since the last one, and posts that match your keywords can go straight to Slack, Discord, any webhook (Make, Zapier, n8n) or your e-mail.

Give it a channel name (durov, @durov) or a link (https://t.me/durov). For every post you get:

• The content: full text with line breaks and emoji, links, hashtags and mentions.
• Media: photos, videos with thumbnail and duration, albums with every picture, files and audio with their names, polls with options and results, link previews with title, description and image.
• Engagement: views, reactions per emoji, paid Telegram Stars.
• Context: date, edited flag, author signature, forwarded-from source, the post it replies to, and the channel's title and subscriber count.

At a glance

	This Actor
Date range	✅ exact: jumps straight to the end of your range and stops reading at its start
Only new posts since the last run	✅ built in, nothing skipped, nothing twice
Keyword filter with alerts to Slack, Discord, webhook or e-mail	✅ built in, no extra integration needed
Channel that cannot be read (group, bot, private, misspelled)	✅ named in the run status, with the reason in the log
Price	✅ $0.50 per 1,000 posts, no start fee; media and posts filtered out cost nothing extra

Why use it?

• Exact date ranges, fast. Set Posted from and Posted until and the Actor finds the right spot in the channel in a few requests, even in channels with hundreds of thousands of posts. Reading stops as soon as older posts begin, so you only pay for posts in the range.
• Monitoring that never misses a post. With Only new posts, each run continues exactly after the last post it delivered. If a run stops early (spending limit, Telegram not answering), the next run picks up from there.
• Keyword alerts built in. Add keywords, a Slack, Discord, webhook or e-mail target, and a schedule. Each alert lists the new matching posts with a link, and an alert that could not be delivered is sent again with the next run.
• Clear answers. If an entry is a group, a bot, a personal account, an invite link or a channel that does not exist, the run says so instead of returning nothing.
• Light and reliable. The Actor reads Telegram's public web preview directly. It needs no proxy, retries hiccups by itself, and ends a run cleanly if Telegram stops answering, so no time or credit is spent waiting.
• Fits your workflow. Schedule it, export to JSON, CSV, Excel or Google Sheets, or call it from Make, Zapier or the API.

Who uses it

• Crypto and trading teams following signal, news and project channels.
• Brand and reputation monitoring: get told when your company, product or domain is mentioned.
• Researchers, journalists and OSINT analysts archiving channels and analysing reach and topics.
• Marketing teams comparing competitors' channels: posting frequency, views and reactions.
• AI and data teams collecting text for search, classification or RAG.

How to use it

1. Enter one or more channels: durov, @durov, https://t.me/durov or a link to any post of the channel.
2. Optionally set Posted from / Posted until (a date or a relative time such as 7 days) and Keywords.
3. Set Max posts per channel and click Start.
4. Download the posts from the Output tab, or read them through the API.

To monitor channels: switch on Only new posts, optionally add keywords and a Slack, Discord, webhook or e-mail target, save the input as a task, and give the task a schedule (for example every hour). The first run returns the latest posts as usual and marks the starting point, without alerts. Every later run returns, and alerts about, only posts published since then. Give the monitor a task or a Monitor name so you can change keywords and alert targets later without starting over.

How much does it cost to scrape Telegram channels?

Pay per event: $0.50 per 1,000 posts returned ($0.0005 per post). Apify platform usage is included, there is no start fee, and posts that your date range or keywords filter out are not charged. E-mail alerts are sent by Apify's own mail Actor, which adds about $0.001 of platform usage per e-mail; Slack, Discord and webhook alerts are free.

What you do	Price
Last 100 posts of 10 channels	$0.50
A year of a busy news channel (about 30,000 posts)	$15
Hourly keyword monitoring of 20 channels with 50 matches a day	about $0.75 per month

Free Apify plan users get $5 of credit every month, which is 10,000 posts.

Input

Field	What it does
channels	Channel usernames or t.me links.
maxPostsPerChannel	Posts returned per channel, newest first. In monitoring mode: new posts per run, oldest first; any others follow with the next run.
dateFrom, dateTo	Only posts published in this period (UTC). A day (2026-10-01), a full time (2026-10-01T08:00:00Z) or a relative time (7 days, 12 hours).
keywords	Only posts that contain at least one of them, in the text, linked addresses, link preview, file names or poll. Case does not matter; bitcoin also finds Bitcoins. Put a keyword in quotes ("AI") to match the exact word only. Without Posted from, a keyword search stops after looking through about 10,000 posts of a channel; monitoring runs read every new post.
excludeKeywords	Skip posts that contain any of these.
onlyNewPosts, monitorName	Monitoring: return only posts published since the previous run.
slackWebhookUrl, discordWebhookUrl, webhookUrl, notifyEmail	Where to send alerts about new posts. E-mail alerts go to the e-mail address of your Apify account.
proxyConfiguration	Usually not needed: the Actor reads Telegram directly. A proxy set here is used for every request.

Example:

{
  "channels": ["cointelegraph", "https://t.me/durov"],
  "keywords": ["bitcoin", "\"ETF\""],
  "onlyNewPosts": true,
  "slackWebhookUrl": "https://hooks.slack.com/services/…"
}

Output

One item per post:

{
  "channel": "cointelegraph",
  "channelTitle": "Cointelegraph",
  "channelSubscribers": 329000,
  "postId": 72796,
  "url": "https://t.me/cointelegraph/72796",
  "date": "2026-10-22T05:49:52Z",
  "edited": false,
  "author": null,
  "type": "photo",
  "text": "🚨 UPDATE: Zcash sets its Network Upgrade 7 to go live on mainnet at block 3,543,405, expected around Nov. 5.\n\nNews | Markets | YouTube",
  "views": 4300,
  "reactions": [
    { "emoji": "❤", "customEmojiId": null, "count": 25 },
    { "emoji": "🔥", "customEmojiId": null, "count": 11 }
  ],
  "paidStars": null,
  "media": [
    { "type": "photo", "url": "https://cdn5.telesco.pe/file/TSr3mz….jpg", "postUrl": "https://t.me/cointelegraph/72796" }
  ],
  "linkPreview": null,
  "poll": null,
  "forwardedFrom": null,
  "replyTo": null,
  "links": ["https://cointelegraph.com/", "https://cointelegraph.com/category/markets", "https://www.youtube.com/@cointelegraph"],
  "hashtags": [],
  "mentions": [],
  "albumPostIds": null,
  "scrapedAt": "2026-10-22T10:42:18Z"
}

type is text, photo, video, album, document, audio, poll or other. With keywords, each post also has matchedKeywords. Link previews come as {url, siteName, title, description, imageUrl}, polls as {question, type, voters, options: [{text, percent}]}, replies as {url, postId, author, text}.

The webhook alert is a JSON POST with monitor, newPostsCount, runUrl and newPosts: the main fields of up to 1,000 posts per alert, with the first 2,000 characters of each text. The full items are always in the run's dataset. More than 100 new posts are sent in several posts, numbered by part and parts.

Use it through the API

Python:

from apify_client import ApifyClient

client = ApifyClient("<YOUR_API_TOKEN>")
run = client.actor("layerwiselabs/telegram-channel-scraper").call(run_input={
    "channels": ["durov"],
    "dateFrom": "30 days",
    "maxPostsPerChannel": 500,
})
for post in client.dataset(run.default_dataset_id).iterate_items():
    print(post["date"], post["views"], post["text"][:80])

AI agents can call the Actor through the Apify MCP server as a tool, for example "what did these five crypto channels post about Solana this week?".

FAQ

Which channels can it read?

Every public channel whose web preview is on, which is the default: the same posts anyone sees at t.me/s/<channel> without logging in. Groups, private channels and invite links (t.me/+…) need a Telegram account, so the run names them as not readable and carries on with the other channels.

How does monitoring work?

With Only new posts on, the Actor remembers the last post it delivered for each channel, separately for each task or Monitor name. Every later run reads the channel from that post onwards and returns what is new, oldest first. In a task, or with a Monitor name, you can change keywords or alert targets at any time and the monitor carries on from the same post. Runs without either keep a separate memory for each combination of keywords and alert targets, so they never mix. If a run of the same monitor is still reading a channel when the next run starts, the new run leaves that channel to it, so no post is reported twice.

How exact are the view counts?

They are the numbers Telegram shows: exact below 1,000, rounded above (5.83M becomes 5830000).

Do media links stay valid?

Photo and video links point to Telegram's CDN and work right after the run. Telegram renews them from time to time, so download the files you want to keep soon after the run; url of the post always leads back to it.

What happens if Telegram does not answer?

Short hiccups are retried. If Telegram stops answering altogether, the run ends cleanly after a few tries, keeps everything delivered so far, and says which channels to run again. A monitor simply continues where it stopped.

Is it legal to scrape Telegram channels?

The Actor reads only posts that channel owners publish openly on the web. Check that your use complies with the laws that apply to you, including data protection rules for any personal data in posts, and with Telegram's terms.

Something does not work

Open an issue in the Issues tab with the input you used, and we will look into it.