YouTube Scraper: extract videos, channels, comments and transcripts

What are the YouTube Scraper specifications?

YouTube Scraper accepts 5 URL types, @handles, keyword searches and a trending switch, shaped by 6 filters. Each video returns metadata, engagement counts, a channel object, thumbnails, streaming formats and the caption track list, exported as JSON or CSV.

What does YouTube Scraper output look like?

YouTube Scraper output is one JSON object per video, carrying the ID, title, URL, duration in seconds, views, likes, publish date, live and family-safe flags, a nested channel object, thumbnails, streaming formats and the caption track list.

{
  "keywords": ["machine learning tutorial"],
  "uploadDate": "w",
  "sort": "v",
  "gl": "us",
  "hl": "en",
  "maxItems": 200
}

{
  "type": "video",
  "id": "wwSzpaTHyS8",
  "title": "Did The Future Already Happen?",
  "duration": 755,
  "views": 6501854,
  "likes": 282499
}

{
  "youtubeHandles": ["@kurzgesagt"],
  "includeShorts": true,
  "maxItems": 300
}

"channel": {
  "id": "UCsXVk37bltHxD1rDPwtNM8Q",
  "name": "Kurzgesagt – In a Nutshell",
  "url": "http://www.youtube.com/@kurzgesagt"
}

{
  "startUrls": [
    "https://www.youtube.com/watch?v=xuCn8ux2gbs",
    "https://www.youtube.com/shorts/vVTa1_hm4n4"
  ],
  "sort": "top",
  "includeReplies": false,
  "maxItems": 1000
}

{
  "text": "This one hit you in the feels? …",
  "likeCount": "3.4K",
  "replyCount": "8",
  "publishedTime": "5 months ago",
  "author": { "id": "UCa90xqK2odw1KV5wHU9WRhg", "name": "@TheOffice" }
}

{
  "startUrls": ["https://www.youtube.com/watch?v=dQw4w9WgXcQ"],
  "lang": "en"
}

{
  "type": "transcript",
  "id": "dQw4w9WgXcQ",
  "transcript": [
    { "text": "Welcome back to the channel…", "dur": "3.52", "startTime": "0" },
    { "text": "Let's take a closer look…", "dur": "2.87", "startTime": "3.6" }
  ],
  "selected": { "title": "English (auto-generated)", "languageCode": "en" }
}

What the fields are

• Video metadata

  ID, title, URL, description, duration in seconds, publish and upload dates, live, private and family-safe flags

• Engagement

  View count and like count

• Channel

  Channel ID, display name and channel URL on every video

• Media and captions

  Thumbnails, embed URL, streaming formats, and the caption track list with language codes

Which Apidojo YouTube scraper fits your job?

Apidojo publishes 7 YouTube scrapers. YouTube Scraper collects videos from channels, playlists, searches, Shorts and trending at $0.50 per 1,000, and six single-job scrapers handle transcripts, comments, channel profiles, channel videos, playlists and trending lists, each billing a small fee per query on top of the same per-video rate.

YouTube scrapers for a single job

• YouTube Transcript Scraperapidojo/youtube-transcript-scraperThe full transcript of a video as timed lines, from a youtube.com or youtu.be watch URL, in the language you set with lang.text, startTime and dur per line · Every available caption language listed · Videos without a transcript are skipped and not charged$0.001 / transcript17 users · new · New
• YouTube Comments Scraperapidojo/youtube-comments-scraperComments from video and Shorts URLs, sorted Top or Latest, with the commenter’s channel ID and name. Reply threads are added with includeReplies.Around 20 comments included per video · 251 comments/sec on videos, 267 on Shorts · Each reply thread billed as its own query$0.001 / video + $0.50 / 1K3K users · 4.8 (12)
• YouTube Channel Info Scraperapidojo/youtube-channel-information-scraperChannel profiles rather than videos, from channel URLs, @handles or a keyword search across channels.Subscriber and video counts, description, tags · Verified, artist and family-safe flags; available countries · 166 channels per second$0.005 / URL · $0.001 / search + $0.50 / 1K2.2K users · 5.0 (8)
• YouTube Trending Scraperapidojo/youtube-trending-scraperThe current trending list for one country, set with a gl country code. The only other input is maxItems.Around 50 videos included per query · One country per query · 14 videos per second$0.025 / query + $0.50 / 1K893 users · 5.0 (7)
• YouTube Channel Videos Scraperapidojo/youtube-channel-scraperThe videos of a channel from its URL or @handle, or channels found by keyword first. Named for what it returns, because its listing shares a title with Channel Info Scraper.Around 30 videos or 48 Shorts included per channel · Around 20 channels included per keyword search · 13 videos per second$0.015 / channel · $0.005 / search + $0.50 / 1K707 users · 5.0 (7)
• YouTube Playlist Scraperapidojo/youtube-playlist-scraperEvery video in a playlist from its URL, or playlists found by keyword, with views, likes and duration per video.Around 100 videos included per playlist · Keyword search with gl, hl and sort · 13 videos per second$0.025 / playlist · $0.005 / search + $0.50 / 1K591 users · 5.0 (7)

How much does YouTube Scraper cost?

YouTube Scraper costs $0.50 per 1,000 videos, billed per result with no per-run and no per-query fee. Scraping 100,000 YouTube videos costs $50.00. The six single-job scrapers add a query fee between $0.001 and $0.025, most of them with items included in it.

Prices apply to Apify paid plans. The free plan runs YouTube Scraper in demo mode, 5 runs a month, 10 items a run, no credit card.

How does YouTube Scraper compare with the YouTube Data API?

The YouTube Data API is free but capped. A Google Cloud project gets 10,000 quota units a day for most endpoints plus a separate bucket of 100 search.list calls, both reset at midnight Pacific Time, and every caption endpoint requires OAuth authorization. YouTube Scraper has no quota, no key and no OAuth step.

Quota figures from Google’s own "Determine quota cost" page, read on the snapshot date, which states that search.list and videos.insert have their own quota buckets with a default daily limit of 100 each, separate from the 10,000 units shared by every other endpoint. That is the number worth knowing: a project can burn its entire search allowance in 100 calls while 10,000 units sit unused.

How to scrape YouTube with Apidojo in 4 steps

Scraping YouTube with Apidojo takes 4 steps: open the scraper, add URLs, handles or keywords, set the filters, then run and export. No YouTube API key, Google Cloud project or OAuth consent screen is involved.

1. 01

   Open YouTube Scraper on Apify

   Sign in with a free Apify account. No credit card is needed to run it in demo mode.

2. 02

   Add URLs, handles or keywords

   Paste a channel, playlist, Shorts or search URL, type an @handle, or enter search keywords.

3. 03

   Set the filters

   Pick country and language, upload date, duration, features and sort; turn on Shorts or trending. Cap the run with maxItems.

4. 04

   Run and export the dataset

   Download JSON or CSV, or read the results straight through the Apify API.

How do you call the YouTube Scraper API?

YouTube Scraper is the actor apidojo/youtube-scraper, called through the Apify REST API or the official Python and JavaScript clients. The input is the same JSON the web form produces, so a run built in the form copies straight into code.

from apify_client import ApifyClient

client = ApifyClient("YOUR_APIFY_TOKEN")

# This week's most-viewed videos for a topic, US results in English
run = client.actor("apidojo/youtube-scraper").call(run_input={
    "keywords": ["machine learning tutorial"],
    "uploadDate": "w",
    "sort": "v",
    "gl": "us",
    "hl": "en",
    "maxItems": 200,
})

for v in client.dataset(run.default_dataset_id).iterate_items():
    print(v["views"], v["duration"], v["title"], v["url"])

import { ApifyClient } from 'apify-client';

const client = new ApifyClient({ token: process.env.APIFY_TOKEN });

// A channel's long-form videos plus its Shorts, split by duration
const run = await client.actor('apidojo/youtube-scraper').call({
  youtubeHandles: ['@kurzgesagt'],
  includeShorts: true,
  maxItems: 300,
});

const { items } = await client.dataset(run.defaultDatasetId).listItems();
const shorts = items.filter((v) => v.duration <= 60);
console.log(shorts.length, 'Shorts of', items.length, 'videos');

curl -X POST \
  "https://api.apify.com/v2/acts/apidojo~youtube-scraper/run-sync-get-dataset-items?token=YOUR_APIFY_TOKEN" \
  -H "Content-Type: application/json" \
  -d '{"startUrls":["https://www.youtube.com/playlist?list=PLrAXtmErZgOeiKm4sgNOknGvNjby9efdf"],"maxItems":100}'

from apify_client import ApifyClient

client = ApifyClient("YOUR_APIFY_TOKEN")

# Transcripts are a separate actor, the main scraper returns the track list
run = client.actor("apidojo/youtube-transcript-scraper").call(run_input={
    "startUrls": ["https://www.youtube.com/watch?v=_AbFXuGDRTs"],
    "lang": "en",
})

for item in client.dataset(run.default_dataset_id).iterate_items():
    for line in item["transcript"]:
        print(line["startTime"], line["text"])

• n8n

  A daily trending run feeding a market digest.

• Make and Zapier

  New channel video into a review queue.

• Warehouse

  JSONL into BigQuery or Snowflake.

• Webhooks

  POST to any URL on run success.

How fast is YouTube Scraper?

YouTube Scraper returns 19 videos a second on search URLs, 16 on channels, 14 on playlists and 10 on keyword searches, all from its own published benchmark. At the channel rate a 10,000-video export finishes in about ten minutes.

Every rate is the figure its own listing publishes, and the transcript row says so rather than borrowing a neighbour’s number. The bars compare the four query types of one actor, because those share a unit; the table mixes videos, comments and channels, which do not. Real throughput moves with query type, region, time of day and how many deleted or private videos get skipped.

30.2K

users on YouTube Scraper

6.5M

YouTube Scraper runs, all time

4.4 / 5

across 63 reviews, seven scrapers

99.6%

run success, 30 days

The 4.4 is weighted by review count across all seven, which pulls it down rather than up: YouTube Scraper’s own rating is 3.3 over 22 reviews, the lowest in the catalogue, while the six single-job scrapers sit between 4.8 and 5.0. Publishing the blend rather than the siblings’ average is the point, the rating you see includes the scraper this page leads with.

What can you build with YouTube Scraper data?

YouTube Scraper data powers 6 workflows: timestamped transcript search, launch-video comment mining, creator shortlisting, per-country trending reports, Shorts-versus-long-form analysis and course audits. Four of the six chain two scrapers together.

• 01

  Timestamped transcript search

  keywords → video URLs → Transcript Scraper

  Fields

  transcript[].text, startTime, url

  Lands in

  A RAG index that cites the exact second a claim was made

• 02

  Launch-video comment mining

  Comments Scraper · sort Latest · includeReplies

  Fields

  text, likeCount, replyCount

  Lands in

  Clustered questions and complaints for product teams

• 03

  Creator shortlisting

  Channel Info Scraper · keywords · gl

  Fields

  subscriberCount, videosCount, availableCountries, tags

  Lands in

  A sponsorship shortlist by niche and market

• 04

  Per-country trending reports

  Trending Scraper · one run per gl, daily

  Fields

  title, views, channel

  Lands in

  A daily market snapshot at $0.75 a month per country

• 05

  Shorts versus long-form

  YouTube Scraper · @handle · includeShorts

  Fields

  duration, views, publishDate

  Lands in

  Views per format for the same channel

• 06

  Course and series audits

  Playlist Scraper · playlist URL

  Fields

  duration, views, likes per video

  Lands in

  Total runtime and viewer drop-off across a series

What are the limits of YouTube Scraper?

YouTube Scraper operates under 6 documented limits. Each one below states what triggers it and how to work within it.

• At least 10 videos per input

  Each keyword, handle or start URL must fetch 10 videos or more. Smaller requests are too costly to serve and can trip the rate limiter.

  Set maxItems to 10 or more and group small inputs into one run.

• Single video URLs are not supported

  The README states this directly, and it follows from the 10-video minimum. The input schema still accepts the watch-URL format, so a single-video run fails rather than being rejected up front.

  Batch at least ten videos, or use Comments Scraper or Transcript Scraper, both of which take one watch URL.

• Five inputs per run, one run at a time

  The README asks for no more than 5 keywords, handles or URLs in a run, at most one concurrent run, and a couple of minutes between runs.

  Queue runs on a schedule rather than firing them in parallel.

• uploadDate and sort are search-only

  Both are YouTube search parameters, so they are ignored on channel, playlist and URL runs.

  Search the channel name as a keyword, or filter on publishDate after the run.

• No caption text, and no comment counts

  This scraper returns the caption track list, not the transcript, and its output carries no comment count. Both come from sibling actors.

  Transcript Scraper for the text at $0.001 a video; Comments Scraper for comments at $0.001 a video.

• Private, member-only and restricted videos

  Private, member-only and deleted videos are not returned at all. Age-restricted videos usually come back without streamingData, and geo-blocked videos depend on where the run executes.

  Open the URL in a logged-out browser first, and plan for empty media fields on restricted content.

Is it legal to scrape YouTube?

Collecting publicly visible YouTube data without logging in is generally lawful in the United States. US courts have held that scraping public pages does not breach the Computer Fraud and Abuse Act, in hiQ Labs v. LinkedIn (9th Cir., 2022). Personal data in comments and channel profiles remains subject to GDPR and CCPA.

• What the scrapers read

  Public, logged-out content only: videos, channels, playlists, trending lists, comments and published captions. No authentication is bypassed and no private content is reached.

• Commenters are people

  A comment author’s name and channel ID identify someone under GDPR. Processing them needs a lawful basis, a stated purpose and a retention period, the same as any personal data.

• Videos and transcripts are creators’ work

  streamingData links to the video files themselves, and a transcript reproduces what was said. Analysing either is one thing; republishing needs the rights holder’s permission.

This is general information, not legal advice. The actor’s own README asks that results be used in accordance with applicable law and YouTube’s Terms of Service.

YouTube Scraper frequently asked questions