YouTube Comments Scraper: With Replies

Get the comments of YouTube videos with their replies, threaded: each reply names the comment it answers. Every row also carries the video's title, views, likes, date and channel, so one file holds the comments and what they are about. See which comments the channel pinned, hearted or replied to. No YouTube account, login or cookies needed.

Tested daily. Latest checks: 132 comments and replies from two videos in 13 seconds; 654 from three videos, each read to its end, in about 3.5 minutes.

What does this Actor do?

You give it links to YouTube videos (or just video ids). For each video it reads the comments, and the replies under each comment, and saves one row per comment or reply until it reaches your limit or the end of the comments.

Each row comes with:

• the comment's text, when it was posted (as YouTube shows it, such as "3 weeks ago"), whether it was edited, its likes and its number of replies
• for a reply: the comment it answers
• whether the channel pinned it, hearted it or replied to it
• the links in the comment, with YouTube's redirect removed
• the commenter's channel id, name, handle and picture, and whether they are verified, the video's own channel, or an official artist
• the video's title, description, views, likes, date, comment count and whether it is live
• the channel's id, name, handle, subscriber count and verified mark

Why use it?

• Replies included, threaded. Replies are read by default, and each one links back to its comment, so you can rebuild every conversation.
• Comments with their context. Each row names its video and channel, with views and likes, so comments from many videos can be compared in one sheet.
• Find what the channel noticed. See which comments it pinned, hearted or answered.
• Top or newest. Read comments in YouTube's top order or newest first.
• Use the data anywhere. Download as JSON, CSV or Excel, or send it to your own tools through Apify's API and integrations.

How to use it

1. Paste one or more video links into Video links.
2. Optionally, set Max comments per video, choose Order, and turn Include replies off.
3. Click Start. When the run ends, open the Output tab to view the comments or download them.

Run it with no input to see a sample: up to 100 comments each from two videos, charged as usual.

Links that work

A video link as in your browser's address bar (https://www.youtube.com/watch?v=IdPutKgWVCE), a youtu.be/… share link, a Shorts, live or embed link, or the 11-character video id on its own (IdPutKgWVCE). Channel, playlist and search pages are not supported: this Actor reads the comments of the videos you name.

Use it with YouTube Scraper's results

Already have video links in a YouTube Scraper run? Skip the copying:

1. Run the YouTube Scraper, then open its run and copy the dataset id from the Storage tab (or paste the dataset's link).
2. Paste it into Dataset id from a scraper run (datasetId) here. Or schedule this Actor after the other one and pass the id each time. If your account cannot read that dataset, paste its rows into Rows from a scraper run (datasetRows) instead.
3. Click Start. The Actor reads each row's url as a link, adds it to Video links and reads its comments. Our YouTube Search and YouTube Channel Videos Actors write the same url field, so their datasets work too.

Rows with no usable link, such as channel or playlist rows, are skipped, and so are rows whose own comment count is 0. The status line says how many rows were read and how many were skipped. A link in both Video links and the dataset is read once. If Video links still holds the form's example links when you give a dataset, they are left out, and the status says so. At most 10,000 rows are read from one dataset. Comments are priced as usual: each comment saved is one row, whichever way the links reached us.

Input

Field	What it does	Default
Video links (startUrls)	The videos to read. Required.	—
Max comments per video (maxCommentsPerVideo)	Stop after this many rows per video, replies included	100
Include replies (includeReplies)	Also save the replies to each comment	true
Order (sortBy)	top (YouTube's default) or newest	top

Example input:

{
  "startUrls": [{ "url": "https://www.youtube.com/watch?v=IdPutKgWVCE" }],
  "maxCommentsPerVideo": 1000,
  "includeReplies": true,
  "sortBy": "newest"
}

Output

One row per comment or reply. An example, with made-up values:

{
  "comment_id": "UgzAbCdEfGhIjKlMnOp4AaABAg",
  "video_id": "IdPutKgWVCE",
  "source_url": "https://www.youtube.com/watch?v=IdPutKgWVCE",
  "text": "Great video, thanks!",
  "published_time_text": "3 weeks ago",
  "is_edited": false,
  "like_count": 38,
  "reply_count": 2,
  "is_reply": false,
  "parent_comment_id": null,
  "reply_level": 0,
  "pinned": false,
  "hearted_by_creator": true,
  "creator_replied": false,
  "links": [],
  "comment_url": "https://www.youtube.com/watch?v=IdPutKgWVCE&lc=UgzAbCdEfGhIjKlMnOp4AaABAg",
  "author_channel_id": "UC0000000000000000000000",
  "author_name": "@janedoe",
  "author_handle": "@janedoe",
  "author_avatar_url": "https://yt3.ggpht.com/…",
  "author_is_verified": false,
  "author_is_channel_owner": false,
  "author_is_artist": false,
  "author_badge": null,
  "video_title": "A video title",
  "video_description": "The video's description…",
  "video_view_count": 136219,
  "video_is_live": false,
  "video_like_count": 3546,
  "video_date_text": "Nov 29, 2024",
  "video_published_date": "2024-11-29",
  "video_comment_count": 260,
  "channel_id": "UC1111111111111111111111",
  "channel_name": "A Channel",
  "channel_handle": "@achannel",
  "channel_subscribers_text": "266K subscribers",
  "channel_verified": true
}

Field	What it holds
comment_id	YouTube's id for the comment or reply
video_id	The video's id
source_url	The video link or id you gave
text	The text
published_time_text	When it was posted, as YouTube shows it ("3 weeks ago")
is_edited	Whether it was edited
like_count	Likes (YouTube rounds counts above 1,000, such as 1.2K → 1200)
reply_count	Replies to the comment (empty on a reply)
is_reply	Whether the row is a reply
parent_comment_id	For a reply: the comment it sits under
reply_level	0 for a comment, 1 for a reply
pinned	Whether the channel pinned it (empty on a reply)
hearted_by_creator	Whether the channel hearted it
creator_replied	Whether the channel replied to it (empty on a reply)
links	Links in the text: outside links, hashtags, mentions and timestamps
comment_url	A link to the comment
author_channel_id	The commenter's channel id
author_name	The commenter's name as shown (usually the @ handle)
author_handle	The commenter's @ handle
author_avatar_url	The commenter's picture
author_is_verified	Whether the commenter is verified
author_is_channel_owner	Whether the commenter is the video's own channel
author_is_artist	Whether the commenter is an official artist channel
author_badge	The commenter's badge text, if any
video_title	The video's title
video_description	The video's description
video_view_count	The video's views (empty while it is live)
video_is_live	Whether the video is a live stream right now
video_like_count	The video's likes
video_date_text	The video's date as YouTube shows it ("Premiered Nov 29, 2024")
video_published_date	The same date as YYYY-MM-DD
video_comment_count	The comment count shown above the comments
channel_id	The video's channel id
channel_name	The channel's name
channel_handle	The channel's @ handle
channel_subscribers_text	The channel's subscribers as YouTube shows them ("266K subscribers")
channel_verified	Whether the channel is verified

Not included: an exact date and time for each comment (YouTube shows only "3 weeks ago"), the video's transcript, and the commenters' own channels.

How much does it cost?

No surcharges and no paid options; an item that fails is not charged.

You pay for what you get:

• $0.50 per 1,000 comments ($0.0005 a comment). A reply counts as a comment. Each is charged once per run.
• No start fee, and nothing per video.

So 2,000 comments and replies from one video cost $1. If you set a maximum cost for a run, the Actor stops when it reaches it and keeps every comment saved so far.

How long does a run take?

Expect about 200 comments and replies a minute. In testing, 654 from three videos, each read to its end, took about 3.5 minutes.

Good to know

• You may get fewer comments than the video's count. YouTube's count includes comments it no longer shows (held for review or removed), so a whole video usually reads a little below video_comment_count.
• Dates are relative. YouTube shows when a comment was posted only as "3 weeks ago" or "1 year ago"; the row keeps it as shown.
• Check your links first. A link that is not a YouTube video link stops the run before anything is read, and the message names it.
• One video that cannot be read does not stop the run. The other videos carry on, and the run's status lists the videos that could not be read. The run fails only when no video could be read.
• No comments is not an error. A video with comments turned off saves nothing.

More Actors from us

• YouTube Channel Scraper: Every Video & Short: Every upload of a YouTube channel, newest first: videos, Shorts and live streams in one list.
• YouTube Search Scraper: Videos, Shorts & Channels: YouTube search results in YouTube's own order: videos, Shorts, channels, playlists, movies.
• YouTube Transcript Scraper: Channels, Playlists & SRT: YouTube transcripts with timestamps from videos, whole channels or playlists, in the languages you pick.
• AliExpress Product Scraper: Search Results & Prices: Every product of an AliExpress search to its last page, with exact sold counts and ratings.
• Amazon Product Scraper: Search Results & Prices: Every product on Amazon's search pages to the last page, sponsored marked, with monthly sales.
• App Store Reviews Scraper: All Reviews, Not 500: Every written App Store review of an app, not just the newest 500.
• Woolworths Coles ALDI Scraper: AU Grocery Prices: Woolworths, Coles and ALDI products for any search in one run, every results page to the end.
• Bilibili Video Scraper: Search & Channel Videos: Bilibili videos from searches or whole channels, each with its full detail page's data.
• Career Site Jobs Scraper: Workday, Greenhouse, Lever: Every job on Greenhouse, Lever, Ashby, SmartRecruiters and Workday boards, checked against each board's count.

Use it from an AI agent or the API

One call starts a run and returns its items (Apify waits up to 300 seconds; for a longer run, start it with POST /v2/acts/…/runs and read its dataset):

POST https://api.apify.com/v2/acts/luzivanlt~youtube-comments-scraper/run-sync-get-dataset-items?token=<YOUR_APIFY_TOKEN>

With the sample input as its JSON body:

{
  "startUrls": [
    {
      "url": "https://www.youtube.com/watch?v=IdPutKgWVCE"
    },
    {
      "url": "https://www.youtube.com/watch?v=lFn5dqNy7uU"
    }
  ],
  "maxCommentsPerVideo": 100,
  "includeReplies": true,
  "sortBy": "top"
}

Its inputs, in the order of the input form:

Input	What it takes	Default
startUrls (Video links)	Links to YouTube videos (https://www.youtube.com/watch?v=…, youtu.be/…, Shorts, live and embed links, on www., m. or music.youtube.com, with or without https://), or 11-character video ids such as IdPutKgWVCE.	—
datasetId (Dataset id from a scraper run)	The id of a run's dataset from YouTube Scraper (streamers/youtube-scraper), so you do not copy video links by hand.	—
datasetRows (Rows from a scraper run)	Rows pasted from YouTube Scraper (streamers/youtube-scraper) results, as a JSON list, when its dataset id is not readable by your account.	—
maxCommentsPerVideo (Max comments per video)	Stop reading a video after this many comments, replies included.	100
includeReplies (Include replies)	Also save the replies to each comment, each marked with the comment it answers.	true
sortBy (Order)	Top comments first (YouTube's default), or newest first.	"top"

Apify's MCP server, https://mcp.apify.com, lists it as luzivanlt/youtube-comments-scraper. To give an AI agent this Actor as a tool, add the server to the agent's MCP config:

{
  "mcpServers": {
    "apify": {
      "url": "https://mcp.apify.com?tools=luzivanlt/youtube-comments-scraper"
    }
  }
}

The client signs in with OAuth, or add "headers": { "Authorization": "Bearer <APIFY_TOKEN>" }.

Charged per comment delivered; the sample input above costs at most $0.10.

FAQ

Do I need a YouTube account or cookies? No. The Actor reads the comments anyone can see under the video.

Can I get the videos of a channel? Not with this Actor: it reads the comments of the videos you give it.

Can I run it from my own code, on a schedule, or with other apps? Yes. Like any Apify Actor, you can start it through the Apify API, run it on a schedule, and connect it to other apps with Apify's integrations and webhooks.

Is it legal to scrape YouTube?

This Actor reads only comments that anyone can see without logging in. The results include commenters' names, handles, channel ids and pictures, which are personal data protected by laws such as the GDPR. Only collect personal data when you have a legitimate reason to, and check YouTube's terms and the laws that apply to you. If you are unsure, ask a lawyer.

Support

Found a problem or missing a feature? Open an issue on the Issues tab and include the run's link.

---

This Actor is not affiliated with, endorsed by or sponsored by YouTube or Google.