📺 YouTube Scraper — fast, complete, residential-grade

Scrape videos, channels, playlists, search results, Shorts, live streams and comments from YouTube — at residential-proxy speed with a Chrome TLS fingerprint.

✨ Why this scraper

• Complete data. Returns the full set: video metadata, channel /about metadata duplicated on every channel item, parsed descriptionLinks, hashtags, commentsCount, isMonetized, full thumbnails array, captions metadata, streamingData (opt-in), plus inline comments and subtitles.
• Fast. ~85 videos/s from a channel, ~210 videos/s from a playlist, 50 watch records in parallel without throttling — benchmarked live.
• Anti-detection by default. curl_cffi with impersonate="chrome" (real Chrome TLS/JA3 fingerprint) and a fresh Apify Residential session ID per HTTP request — every outbound call uses a different residential IP.
• PoToken-proof subtitles. Automatically falls back to YouTube's iOS player API when the WEB caption baseUrl is empty (post-PoToken rollout), so subtitles actually download.
• Channel /about enrichment. Every channel item carries channelDescription, channelDescriptionLinks, channelJoinedDate, channelLocation, channelTotalVideos, channelTotalViews, numberOfSubscribers, inputChannelUrl.
• Comments inline. Walked through /youtubei/v1/next continuation — no separate actor needed.
• Parallel everywhere. Multiple source URLs scrape concurrently. /about and tab listings on a channel fetch in parallel.

🚀 Quick start

1. Paste a YouTube link (video, channel, playlist, search, Shorts) into Start URLs — or type keywords into Search Queries — or drop channel handles (@MrBeast) into YouTube Handles.
2. Click Start.
3. Download results from the Dataset tab in JSON, CSV, Excel, XML or HTML.

The default settings already have a sample video URL pre-filled, so a one-click run returns one full video record immediately.

🎯 What you can scrape

Source URL	What you get
https://www.youtube.com/watch?v=...	Single video: title, description, likes, views, commentsCount, channel info, subscribers, hashtags, descriptionLinks, full thumbnails array, dates, captions metadata, isLive / isMonetized / isFamilySafe, optional streamingData
https://www.youtube.com/shorts/...	Same shape as a normal video
https://www.youtube.com/@handle or /c/Name or /channel/UC...	Channel /about metadata + every requested tab (videos / shorts / streams), each item enriched with channel-level fields
https://www.youtube.com/playlist?list=...	Playlist videos with channelName, channelUrl, viewCount, relative date, thumbnail
https://www.youtube.com/results?search_query=...	Search results with full listing-level fields
https://www.youtube.com/hashtag/...	Hashtag feed
Search Queries input	Same as above, plus you can pick sort order / date / 4K / HD / CC / Live / etc.

⚙️ Input examples

Scrape one video, including subtitles & comments

{
  "startUrls": [
    "https://www.youtube.com/watch?v=dQw4w9WgXcQ"
  ],
  "downloadSubtitles": true,
  "subtitlesLanguage": "en",
  "subtitlesFormat": "srt",
  "downloadComments": true,
  "maxComments": 200
}

Scrape an entire channel (videos + shorts + streams)

{
  "youtubeHandles": ["@MrBeast"],
  "maxResults": 200,
  "maxResultsShorts": 50,
  "maxResultStreams": 10,
  "sortVideosBy": "POPULAR"
}

Search with filters (HD videos uploaded this week, sorted by views)

{
  "searchQueries": ["climate change"],
  "maxResults": 100,
  "sortingOrder": "views",
  "dateFilter": "week",
  "isHD": true
}

Multi-source batch, with a global cap

{
  "startUrls": [
    "https://www.youtube.com/playlist?list=PL...",
    "https://www.youtube.com/@apify/videos",
    "https://www.youtube.com/results?search_query=apify"
  ],
  "maxResults": 50,
  "maxItems": 500,
  "concurrency": 8
}

📤 Output example

{
  "type": "video",
  "id": "wwSzpaTHyS8",
  "title": "Did The Future Already Happen? - The Paradox of Time",
  "url": "https://www.youtube.com/watch?v=wwSzpaTHyS8",

  "description": "Go to https://brilliant.org/nutshell/ to dive deeper...",
  "descriptionLinks": [
    { "text": "https://brilliant.org/nutshell/", "url": "https://brilliant.org/nutshell/" }
  ],
  "hashtags": ["#science"],
  "keywords": ["physics", "time"],
  "category": "Education",

  "views": 14613126,
  "likes": 461357,
  "commentsCount": 18432,
  "commentsTurnedOff": false,

  "date": "2024-01-30",
  "publishDate": "Jan 30, 2024",
  "uploadDate": "Jan 30, 2024",
  "duration": "00:12:35",
  "durationSeconds": 755,

  "thumbnailUrl": "https://i.ytimg.com/vi/wwSzpaTHyS8/maxresdefault.jpg",
  "thumbnails": [
    { "url": "...hqdefault.jpg", "width": 168, "height": 94 },
    { "url": "...maxresdefault.jpg", "width": 1920, "height": 1080 }
  ],
  "embedUrl": "https://www.youtube.com/embed/wwSzpaTHyS8",

  "channel": {
    "id": "UCsXVk37bltHxD1rDPwtNM8Q",
    "name": "Kurzgesagt – In a Nutshell",
    "url": "https://www.youtube.com/@kurzgesagt"
  },
  "numberOfSubscribers": 22500000,

  "isLive": false,
  "isPrivate": false,
  "isUnlisted": false,
  "isFamilySafe": true,
  "isMonetized": true,
  "playabilityStatus": "OK",
  "availableCountries": ["US", "GB", "DE", "..."],

  "captions": {
    "captionTracks": [
      { "base_url": "...", "language_code": "en", "kind": null }
    ]
  },

  "subtitles": [
    { "type": "manual", "language": "en", "format": "srt", "content": "1\n00:00:02,440 --> 00:00:09,520\nDo your past..." }
  ],

  "comments": [
    {
      "id": "Ug...",
      "author": "@kelving420",
      "text": "The fact they can use all of these trademarked characters...",
      "likeCount": 16,
      "replyCount": 16,
      "publishedTime": "3 weeks ago"
    }
  ],

  "fromYTUrl": "https://www.youtube.com/watch?v=wwSzpaTHyS8",
  "status": "OK"
}

When scraping a channel, each item additionally carries:

"channelDescription": "...",
"channelDescriptionLinks": [{"text": "...", "url": "https://..."}],
"channelJoinedDate": "Feb 1, 2015",
"channelLocation": "United Kingdom",
"channelTotalVideos": 418,
"channelTotalViews": "2,473,486,020",
"inputChannelUrl": "https://www.youtube.com/@RickAstleyYT/about"

⚡ Performance (live benchmark)

Workload	Items	Time	Rate
Channel videos (with continuation)	100	1.17s	85 vid/s
Playlist (with continuation)	100	0.47s	210 vid/s
Search results (with continuation)	100	1.49s	41 vid/s
3 listings in parallel	180	1.07s	167 vid/s
8 channels × 200 videos in parallel	1294	2.0s	656 vid/s
500 comments via /next continuation	500	3.3s	150 com/s
Single full /watch record	1	0.81s	—

🧰 Advanced options

• includeStreamingData — attach raw playerResponse.streamingData (mp4 formats, adaptive formats, URLs). Off by default because most format URLs are signature-protected and the payload is large.
• saveSubsToKVS — write subtitles to the key-value store, keep the dataset record slim.
• concurrency — number of in-flight HTTP requests. Default 5; the actor is happy at 16+.
• maxItems — global cross-source cap (e.g. set 1000 to stop the run after 1000 items regardless of how many start URLs).
• sortVideosBy — channel-tab sort: NEWEST, POPULAR, OLDEST.

🌍 Country / language (gl / hl)

YouTube respects the gl=US / hl=en URL parameters for results localization. The actor threads them through to:

• Every HTML fetch (appended as query parameters)
• The Accept-Language header
• The Innertube context.client.gl / hl for /browse, /search, /next, /player continuation calls
• The residential proxy's country routing — picking gl=de means traffic exits from a German residential IP and you get German-localized results.

❓ FAQ

Why is getTrending returning nothing? YouTube removed the public Trending feed for unauthenticated visitors in 2024 — /feed/trending now serves a "Try searching to get started" placeholder. As a workaround, use searchQueries with "sortingOrder": "views" and "dateFilter": "today".

Why is streamingData empty in my output? It's off by default — flip includeStreamingData: true. Note that for most popular videos the playback URLs inside are signature-ciphered and need a JS decipher step (see signatureCipher vs url keys).

Why are some subtitle tracks missing for a video? Either the video has no captions, or YouTube returned an empty body for the WEB caption URL (post-PoToken rollout). The actor auto-falls back to the iOS-client player, which fixes the vast majority of cases.

Why location: null? The video's uploader didn't tag a location. The field exists on YT's microformat but is rarely populated.

Can I scrape dislikes? No — YouTube made dislike counts non-public in 2021.

What's isMonetized based on? The actor checks playerResponse.adPlacements and playerResponse.playerAds. If either is non-empty, the video runs ads and is monetized.

📞 Support

Open an issue on the Issues tab with a sample input that reproduces the problem. We aim to respond same-day for blockers.