YouTube Comments Scraper – Comments & Replies

Export YouTube comments and replies

Download public comments and replies from YouTube videos and Shorts as structured data for JSON, CSV, Excel, Google Sheets, n8n, and research workflows. Add up to 20 inputs and receive one Dataset item per parent comment or reply, with text, native image/GIF attachments, authors, likes, publication labels, reply relationships, pinned status, creator hearts, badges, and direct comment URLs.

Choose Top comments for a ranked sample, Newest first for chronological collection, or set the limit to 0 to traverse all public comment pages YouTube makes available. No YouTube Data API key is required.

Invalid inputs and videos that cannot produce a complete result receive a visible, free videoStatus record. Previously collected comments remain available if a later request fails. No AI analysis or inferred sentiment is added to the original comments.

This is an unofficial Actor and is not affiliated with, endorsed by, or sponsored by YouTube or Google. It collects publicly accessible comments only.

Ready-to-run examples

• Analyze top YouTube comments
• Download all public YouTube comments
• Scrape YouTube comments and replies

For Shorts, use the Shorts input example below. These examples are separate use cases, not different prices or subscription plans.

Common use cases

• Brand, product, and campaign monitoring
• Audience research and creator analysis
• Sentiment, topic, and keyword analysis
• Community feedback and moderation review
• Market research and competitor tracking
• NLP, LLM, and data-enrichment workflows

Input and quick start

Supported URLs

URL type	Example
Standard video	https://www.youtube.com/watch?v=VIDEO_ID
Short link	https://youtu.be/VIDEO_ID
Shorts	https://www.youtube.com/shorts/VIDEO_ID
Embedded video	https://www.youtube.com/embed/VIDEO_ID
Live video page	https://www.youtube.com/live/VIDEO_ID
Bare video ID	YE7VzlLtp-4

For live video pages, the Actor extracts the normal public comments section when available. It does not extract live-chat messages.

Only HTTP(S) YouTube URLs are accepted. A channel URL, a URL from another website, or an invalid ID receives an invalidInput status instead of being silently dropped. Different links to the same video ID are processed once. A bare ID does not identify whether content is a Short; use its /shorts/ URL when sourceType: "short" is required.

The input form accepts a plain list of URLs or IDs, one per item. API callers can send a string array such as "startUrls": ["YE7VzlLtp-4"] or keep the existing object format "startUrls": [{"url": "YE7VzlLtp-4"}]. Both formats reach the same extraction flow.

Quick start

Start with a small sample of up to 20 ranked parent comments. Only startUrls is required; omitted options use the defaults in the reference table.

{
  "startUrls": [
    "https://www.youtube.com/watch?v=YE7VzlLtp-4"
  ],
  "maxCommentsPerVideo": 20,
  "sortBy": "top",
  "includeReplies": false,
  "maxRepliesPerThread": 10
}

Input reference

Field	Type	Default	Behavior
startUrls	array	Required	1–20 public video/Shorts URLs or bare 11-character IDs. The form uses a string list; existing {url: ...} API entries are also accepted. Duplicate IDs are processed once; invalid entries receive free status rows.
maxCommentsPerVideo	integer	100	1–5000 limits combined parent and reply records per video. Free status rows do not count. 0 follows all available public parent-comment pages.
sortBy	string	top	top for YouTube's ranked view or newest for chronological results. Full collection always uses newest.
includeReplies	boolean	false	Expand and return replies under each parent comment. Disable for the fastest parent-only run.
maxRepliesPerThread	integer	10	1–500 replies per parent when includeReplies is enabled, including full-collection mode. Ignored when replies are disabled.

Shorts comments

{
  "startUrls": [
    {"url": "https://www.youtube.com/shorts/cjSKS1D9Kow"}
  ],
  "maxCommentsPerVideo": 20,
  "sortBy": "top",
  "includeReplies": false
}

Comments with replies

{
  "startUrls": [
    {"url": "https://www.youtube.com/watch?v=YE7VzlLtp-4"}
  ],
  "maxCommentsPerVideo": 100,
  "sortBy": "top",
  "includeReplies": true,
  "maxRepliesPerThread": 10
}

The limit of 100 counts parents and replies together. It does not mean 100 parents plus their replies. Replies are collected beneath each parent, so a reply-heavy thread can use a significant part of a small result limit.

Sorting and full collection

Result limit	Requested sort	Effective sort	Result
Positive number	top	top	Fixed-size ranked sample
Positive number	newest	newest	Fixed-size chronological sample
0	top or newest	newest	All available public comment pages

To follow every available public parent-comment page and include up to 500 replies per parent:

{
  "startUrls": [
    {
      "url": "https://www.youtube.com/watch?v=VIDEO_ID"
    }
  ],
  "maxCommentsPerVideo": 0,
  "sortBy": "newest",
  "includeReplies": true,
  "maxRepliesPerThread": 500
}

maxCommentsPerVideo: 0 removes the combined result limit, not the per-thread reply cap, the run's spending limit, or the platform run timeout. “All” means the public pages YouTube exposes to this unauthenticated session—not private, deleted, held-for-review, or otherwise unavailable comments. Top-ranked pagination can finish before the requested number is reached; a maximum is not a guaranteed result count.

Dataset output

Each parent comment and included reply is a separate recordType: "comment" item.

Real comment example

This complete parent-comment record came from a real local run on October 20, 2026. Public counters and relative dates can change.

{
  "recordType": "comment",
  "sourceType": "video",
  "videoId": "YE7VzlLtp-4",
  "videoUrl": "https://www.youtube.com/watch?v=YE7VzlLtp-4",
  "videoTitle": "Big Buck Bunny",
  "videoChannelId": "UCSMOQeBJ2RAnuFungnQOxLg",
  "videoChannelName": "Blender",
  "commentId": "UgyvM3zp3TKN8kEO4Yd4AaABAg",
  "commentUrl": "https://www.youtube.com/watch?v=YE7VzlLtp-4&lc=UgyvM3zp3TKN8kEO4Yd4AaABAg",
  "text": "still looks better than my renders",
  "publishedTimeText": "3 years ago",
  "publishedAtEstimated": "2023-10-20T02:49:36.443192Z",
  "likeCount": 518,
  "likeCountText": "518",
  "replyCount": 7,
  "parentCommentId": null,
  "replyDepth": 0,
  "isReply": false,
  "isPinned": false,
  "isHearted": false,
  "isPaid": false,
  "authorName": "@5MadMovieMakers",
  "authorChannelId": "UCLkZVEXz_qObWr7hAEbS_pQ",
  "authorChannelUrl": "https://www.youtube.com/@5MadMovieMakers",
  "authorAvatarUrl": "https://yt3.ggpht.com/ytc/AIdro_m_HacbRsFWToHwnxwoZRFsLPGdHSq0GCQGcsTkUWV1FXc=s88-c-k-c0x00ffffff-no-rj",
  "authorIsVerified": true,
  "authorIsVideoOwner": false,
  "authorIsMember": false,
  "authorBadges": [
    "Verified"
  ],
  "sortBy": "top",
  "scrapedAt": "2026-10-20T02:49:36.443192Z"
}

Field groups

Group	Fields
Source	recordType, sourceType, videoId, videoUrl, videoTitle, videoChannelId, videoChannelName
Comment	commentId, commentUrl, text, optional attachments, publishedTimeText, publishedAtEstimated
Engagement	likeCount, likeCountText, replyCount, isPinned, isHearted, isPaid, paidAmountText
Reply relationship	parentCommentId, replyDepth, isReply
Author	authorName, authorChannelId, authorChannelUrl, authorAvatarUrl, authorIsVerified, authorIsVideoOwner, authorIsMember, authorBadges
Run context	sortBy, scrapedAt

publishedAtEstimated is calculated from YouTube's relative label such as 3 years ago. It is useful for sorting, but it is not an exact publication timestamp supplied by YouTube. likeCount is normalized from the displayed value; likeCountText preserves the public label.

Reply records

A reply has isReply: true, replyDepth: 1, and its parent's ID in parentCommentId. Every reply is independently written to the Dataset and counts as one successful comment result.

This excerpt is from a real creator reply on a Short:

{
  "recordType": "comment",
  "sourceType": "short",
  "videoId": "cjSKS1D9Kow",
  "commentId": "UgxGGjnWNMgVP1KmZnZ4AaABAg.AUH2umSNOznAUH8C_g1c9z",
  "text": "you're so sweet!",
  "likeCount": 10,
  "parentCommentId": "UgxGGjnWNMgVP1KmZnZ4AaABAg",
  "replyDepth": 1,
  "isReply": true,
  "authorName": "@sabrina_ramonov",
  "authorIsVideoOwner": true,
  "commentUrl": "https://www.youtube.com/watch?v=cjSKS1D9Kow&lc=UgxGGjnWNMgVP1KmZnZ4AaABAg.AUH2umSNOznAUH8C_g1c9z"
}

Image and GIF comments

An image-only comment can legitimately have text: "". Its attachment is preserved instead of inventing text or dropping the comment. This real-record excerpt shows an animated image returned by YouTube:

{
  "recordType": "comment",
  "videoId": "9bZkp7q19f0",
  "commentId": "UgyugAYtlMUjAmOy_1Z4AaABAg",
  "text": "",
  "attachments": [
    {
      "url": "https://i.ytimg.com/c_exp/UgyugAYtlMUjAmOy_1Z4AaABAg/IMSstM98j7Zgs_webp_animated_original_aspect_ratio.webp?sqp=CPzf2tYGGOCoAQ==&rs=AOn4CLA1cE1O1u6u9MfwHlmr1KJaXUeSTA",
      "width": 498,
      "height": 280,
      "format": "IMAGE_FORMAT_ANIMATED_WEBP"
    }
  ]
}

url, dimensions, and format come from YouTube's response. The original format hint is preserved: a GIF-style animation can be served as animated WebP. Attachment URLs can expire. The Actor does not download, host, or convert these files, and no additional event is charged for the attachment.

Output table

Comments overview keeps the default table compact and includes attachments so image-only comments are not mistaken for missing content. Full comment and status fields exposes author details, badges, reply relationships, and status details. Views change the displayed columns, not the underlying records or available export data. Successful comment rows contain data rather than per-request diagnostics.

Videos without comment rows

Invalid inputs, empty or disabled comment sections, unavailable videos, partial extractions, failures, and budget stops produce free status records. This example is from the real comments-disabled video:

{
  "recordType": "videoStatus",
  "status": "commentsDisabled",
  "inputValue": "https://www.youtube.com/watch?v=XqZsoesa55w",
  "sourceType": "video",
  "videoId": "XqZsoesa55w",
  "videoUrl": "https://www.youtube.com/watch?v=XqZsoesa55w",
  "commentCount": 0,
  "parentCommentCount": 0,
  "replyCommentCount": 0,
  "truncated": false,
  "errorCode": "COMMENTS_DISABLED",
  "errorMessage": "Comments are turned off. Learn more",
  "suggestion": "Use another public video with comments enabled."
}

The example above omits its run timestamp and sort setting for readability.

Status	Meaning
invalidInput	The entry is not a supported video/Shorts URL or ID. The original input is retained, without an invented video ID.
noComments	No public comments were returned, without an explicit comments-disabled message.
commentsDisabled	YouTube explicitly reports that comments are turned off or unavailable.
unavailable	The video is not accessible to a public session.
partial	Some comments were saved, but collection could not finish. truncated is true.
failed	Extraction failed before any comments were saved.
chargeLimitReached	The user-defined spending limit prevented further paid comment records.

Invalid entries do not prevent valid entries from running. An all-invalid input list produces status records and a run summary instead of a traceback. A budget or network failure does not erase comments already saved. Status rows are not comments and never trigger the youtube-comment-result billing event; an Actor start event can still apply.

Pricing

This Actor uses pay-per-event pricing with platform usage included:

• Actor start: $0.00005 per start event; one event per GB of Actor memory, with a minimum of one.
• Comment or reply result: $0.0005 each—$0.50 per 1,000 results.
• Video status row: free.

Example event charges at the default 512 MB memory, before any configured discounts:

Comment and reply records	Event charge including one start
1	$0.00055
100	$0.05005
1,000	$0.50005

The Actor respects the user's maximum run charge. If the budget cannot cover another comment result, collection stops and the input receives a free chargeLimitReached status. The Pricing tab is the source of truth for current prices and discounts.

API and run information

Run summary

The Dataset is the primary output. The default Key-Value Store record named OUTPUT is a compact run report for automation and troubleshooting. It contains raw input totals, invalid input and duplicate counts, comment totals, and per-source results. Duplicate valid IDs are counted but not fetched again.

Use it when a source produced no comment records and you need to distinguish “no public comments” from “request failed.”

Per-video stopReason explains why collection ended, for example endOfPublicPages, maxCommentsReached, maxRepliesPerThreadReached, maxChargeReached, requestBlocked, or networkError. Reaching a configured result/reply cap can produce status: "complete" with truncated: true: the requested bounded job completed, not the video's entire comment history. These collection details stay in the summary rather than being repeated on every comment.

Run through the Apify API

For small runs, the synchronous endpoint returns Dataset items directly:

curl -X POST \
  "https://api.apify.com/v2/acts/arjun_code~youtube-comments-scraper/run-sync-get-dataset-items" \
  -H "Authorization: Bearer $APIFY_TOKEN" \
  -H "Content-Type: application/json" \
  -d '{
    "startUrls": [{"url": "https://www.youtube.com/watch?v=YE7VzlLtp-4"}],
    "maxCommentsPerVideo": 100,
    "sortBy": "top",
    "includeReplies": false,
    "maxRepliesPerThread": 10
  }'

For large full-collection runs, start the Actor asynchronously and read the default Dataset after completion.

Performance guidance

• Start with 10–100 comments when evaluating a source.
• Leave replies disabled if you only need parent comments.
• Use 0 only when complete public pagination is required; popular videos can contain many pages.
• Up to three videos are processed concurrently. Pages for one video remain sequential because YouTube continuation tokens must be followed in order.
• Records are written in small batches, including reply mode. Temporary page-request failures are retried once on that page before an outer retry is considered.
• Full collection can legitimately take longer than a small sample. There is no fixed 40-second deadline for an entire video; the platform run timeout and the user's spending limit still apply.

For context, October 20, 2026 cloud measurements of version 0.1.11 returned 100 parent comments in approximately 8–15 seconds and 1,000 newest parent comments in approximately 55 seconds. These are individual historical samples, not a benchmark or guarantee for this update. Replies, source availability, and network conditions can substantially change runtime.

Troubleshooting

The Dataset contains only a status row

Read status, errorMessage, and suggestion. A video may genuinely have no comments, have comments disabled, be unavailable to a public session, or encounter a temporary request failure.

The collected total is lower than YouTube's displayed count

Displayed totals can include deleted, hidden, moderated, unavailable, or non-public comments. Replies are also capped by maxRepliesPerThread. The Actor stops when YouTube no longer provides a public continuation token.

includeReplies is false—does maxRepliesPerThread matter?

No. The setting is ignored until replies are enabled.

A comment has no text

Inspect attachments in the full-field view or JSON export. YouTube supports image/GIF-style comments that do not contain text. Missing text is not filled with a generated description.

The run is marked successful, but a source returned a partial result

The run can finish normally after saving available data and a status row. Check each source's status, truncated, and the OUTPUT report rather than treating the process status as proof that every source was complete.

Related Actors

Goal	Actor
Discover videos, Shorts, or channels by keyword	YouTube Search Scraper
Get metadata for known video or Shorts URLs or IDs	YouTube Video and Shorts Details Scraper
Enrich known channel handles, IDs, or URLs	YouTube Channel Details Scraper
Export a channel's videos, Shorts, and streams	YouTube Channel Content Scraper

FAQ

Does Top comments return every comment?

No. It is YouTube's ranked view. Set maxCommentsPerVideo to 0; the Actor automatically uses Newest first for the most complete public collection.

Does this support Shorts?

Yes. Supply a /shorts/VIDEO_ID URL. The output uses sourceType: "short".

Does this require a YouTube API key?

No.

Can it scrape private comments or live chat?

No. It returns comments visible to a public, unauthenticated visitor. It does not access private data, creator dashboards, held-for-review comments, or live chat.

Support and responsible use

For unexpected results, open the Actor's Issues tab and include the run ID, public video URL, input settings, expected behavior, and relevant status or log message. Do not post API tokens or private data.

Use the Actor only for lawful processing of publicly accessible information. Follow applicable laws, privacy requirements, and YouTube's terms. Do not use the output for spam, harassment, or invasive profiling.