YouTube Comments Scraper - Comments & Replies, No Login
Pricing
from $0.17 / 1,000 comments
YouTube Comments Scraper - Comments & Replies, No Login
Scrape comments and replies from any public YouTube video or Short: text, likes as numbers, reply counts, pinned and hearted flags, author channel IDs and handles, Top or Newest. $0.25 per 1,000 comments, platform usage included, no per-video fee, errors and duplicates free.
Pricing
from $0.17 / 1,000 comments
Rating
0.0
(0)
Developer
Ragnum
Maintained by CommunityActor stats
0
Bookmarked
2
Total users
1
Monthly active users
9 days ago
Last modified
Categories
Features
Share
YouTube Comments Scraper — Comments & Replies, No Login
Get the comments and replies of any public YouTube video or Short as clean rows: comment text, likes as real numbers, reply counts, pinned and creator-hearted flags, edit status, and the author's channel ID, @handle, name and badges. Choose Top or Newest order. No login, no API key, no account.
$0.25 per 1,000 comments (replies included), platform usage included. No per-video fee and no minimum. Failed videos, videos with comments turned off, and duplicates are free.
What you can do with it
- Audience research: read what viewers say about a product launch, a trailer or a competitor's review video.
- Sentiment analysis and AI: feed thousands of comments into an LLM or a sentiment model; every row has the same keys and typed values.
- Monitor new comments: schedule a run with
"sort": "newest"and a small per-video limit to catch fresh comments on your own videos. - Find engaged fans and creators: sort by
likeCount, filterisHeartedByCreator, or group byauthor.channelId. - Shorts too: paste
https://www.youtube.com/shorts/…URLs like any other video.
Copy to your AI assistant
Use the Apify Actor "ragnum/youtube-comments-scraper" to get YouTube comments.Minimal input: {"videos": ["https://www.youtube.com/watch?v=dQw4w9WgXcQ"], "maxCommentsPerVideo": 100}Options: "sort": "top" | "newest"; "includeReplies": true; "maxRepliesPerComment": 20; "maxResults": 1000.Price: $0.25 per 1,000 comments or replies saved. Error rows (errorCode set) are free.Docs: https://apify.com/ragnum/youtube-comments-scraper.mdMCP: https://mcp.apify.com?tools=ragnum/youtube-comments-scraper
Quick start
100 top comments of one video (the default)
{"videos": ["https://www.youtube.com/watch?v=dQw4w9WgXcQ"]}
Several videos and Shorts, newest comments first
{"videos": ["https://www.youtube.com/watch?v=dQw4w9WgXcQ","https://youtu.be/jNQXAC9IVRw","https://www.youtube.com/shorts/ai1gIWacf0I"],"sort": "newest","maxCommentsPerVideo": 200}
Comments with their replies, capped for a predictable cost
{"videos": ["dQw4w9WgXcQ"],"maxCommentsPerVideo": 500,"includeReplies": true,"maxRepliesPerComment": 10,"maxResults": 3000}
Every comment of a video (can be millions on popular videos; set maxResults as a safety cap)
{"videos": ["https://www.youtube.com/watch?v=dQw4w9WgXcQ"],"maxCommentsPerVideo": 0,"maxResults": 50000}
Input
| Field | Type | Default | What it does |
|---|---|---|---|
videos | array of strings | sample video | Video or Shorts URLs (watch?v=, youtu.be/, /shorts/, /live/, /embed/) or 11-character video IDs. The same video given twice is scraped once. Channel, playlist and search links get a free INVALID_INPUT error row. |
sort | top / newest | top | Top comments (YouTube's default order) or newest first. |
maxCommentsPerVideo | integer | 100 | Top-level comments per video. 0 = all. Replies are not counted here. |
includeReplies | boolean | false | Also save the replies under each comment, one row per reply. |
maxRepliesPerComment | integer | 20 | Replies per comment when includeReplies is on. 0 = all. |
maxResults | integer | none | Cap on all rows (comments + replies) for the whole run. |
hl | string | en | Interface language for texts like "3 days ago" and "1.2K" (comments stay in their own language). |
gl | string | US | Country code for YouTube's regional settings. |
maxConcurrency | integer | 10 | Parallel requests (1–20). Faster for many videos; same price. |
The other advanced settings are managed for you; keep their defaults.
How the limits work together: maxCommentsPerVideo counts only top-level comments. Each of them can bring up to maxRepliesPerComment replies, so one video returns at most maxCommentsPerVideo × (1 + maxRepliesPerComment) rows. maxResults caps the whole run, replies included. Example: 100 comments × (1 + 20 replies) = up to 2,100 rows per video; set maxResults to stay within a budget.
Output
One row per comment or reply. Every row has every field; values YouTube doesn't show are null. Dataset views: Overview and All fields.
| Field | Type | Description |
|---|---|---|
commentId | string | YouTube comment ID. Replies look like <parentCommentId>.<replyId>. |
videoId | string | 11-character video ID. |
videoUrl | string | Video URL (the Shorts URL when you gave a Short). |
videoTitle | string | Video title. |
text | string | Comment text. |
likeCount | number | Likes as a number (0 when none). |
likeCountText | string | Likes as YouTube shows them, e.g. "570K". |
replyCount | number | Replies to this comment (0 for replies). |
publishedTimeText | string | Relative time as shown, e.g. "6 years ago". |
publishedAtEstimate | string | Estimated posting time, ISO 8601 UTC (scrape time minus the relative time). |
isEdited | boolean | The comment was edited (the "(edited)" marker is removed from publishedTimeText). |
isPinned | boolean | Pinned by the creator. |
isHeartedByCreator | boolean | The creator gave it a heart. |
isReply | boolean | true for replies. |
parentCommentId | string | For replies: the comment it answers. null for top-level comments. |
author.channelId | string | Author's channel ID (UC…). |
author.handle | string | Author's @handle. |
author.name | string | Author's display name. |
author.avatarUrl | string | Author's avatar image URL (expires, see Limits). |
author.isVerified | boolean | Verified badge. |
author.isCreator | boolean | The author is the video's creator. |
author.isMember | boolean | Channel membership badge. |
source | string | The input that produced the row. |
scrapedAt | string | When the row was scraped (ISO 8601 UTC). |
Example row
{"commentId": "UgyEnXfdC-umwvTt8JF4AaABAg","videoId": "dQw4w9WgXcQ","videoUrl": "https://www.youtube.com/watch?v=dQw4w9WgXcQ","videoTitle": "Rick Astley - Never Gonna Give You Up (Official Video) (4K Remaster)","text": "Gonna flag this for nudity so I can rick roll the YouTube staff","likeCount": 570000,"likeCountText": "570K","replyCount": 732,"publishedTimeText": "6 years ago","publishedAtEstimate": "2020-11-07T22:00:19.808Z","isEdited": false,"isPinned": false,"isHeartedByCreator": false,"isReply": false,"parentCommentId": null,"author": {"channelId": "UCWf34D_s8K_m2vwBISIDWtg","handle": "@Oatman69","name": "@Oatman69","avatarUrl": "https://yt3.ggpht.com/ghenbV7T5VMOA3iqp3PThC82exqcu7iVng_iWNx1Ujak72Ti4oZZ_XzSVzIrfk9arP1XrFtnZA=s88-c-k-c0x00ffffff-no-rj","isVerified": false,"isCreator": false,"isMember": false},"source": "https://www.youtube.com/watch?v=dQw4w9WgXcQ","scrapedAt": "2026-11-08T10:00:19.808Z"}
Error rows (free)
When a video can't give comments, you get one row for it instead of silence, so automations can match every input to an output. Error rows are never charged.
{"input": "https://www.youtube.com/watch?v=XqZsoesa55w","error": "Comments are turned off for this video.","errorCode": "COMMENTS_DISABLED","retryable": false,"videoId": "XqZsoesa55w","videoUrl": "https://www.youtube.com/watch?v=XqZsoesa55w","videoTitle": "Baby Shark Dance | #babyshark Most Viewed Video | Animal Songs | PINKFONG Songs for Children","source": "https://www.youtube.com/watch?v=XqZsoesa55w","scrapedAt": "2026-11-08T10:00:00.000Z"}
errorCode | Meaning | retryable |
|---|---|---|
COMMENTS_DISABLED | Comments are turned off for this video. | no |
VIDEO_UNAVAILABLE | Private, removed or otherwise unavailable video. | no |
INVALID_INPUT | Not a video URL or ID (e.g. a channel or playlist link). | no |
BLOCKED | YouTube kept refusing the request after several retries. | yes |
HTTP_ERROR | A request failed after retries. If some comments were already saved, commentsSaved says how many. | usually |
To get only comments, filter out rows that have an errorCode.
Pricing
Pay per event, with platform usage included:
| Event | Price | When |
|---|---|---|
comment | $0.00025 ($0.25 per 1,000) | A comment or a reply is saved to the dataset |
| You get | You pay |
|---|---|
| 100 comments | $0.025 |
| 1,000 comments | $0.25 |
| 10,000 comments (e.g. 100 videos × 100) | $2.50 |
- No per-video fee, no start fee, no minimum. One video with 5 comments costs $0.00125.
- Never charged: error rows (comments off, unavailable videos, invalid inputs, failed requests), duplicates, retries.
- Replies cost the same as comments. Use
maxRepliesPerCommentandmaxResultsto control the total. - Set a maximum total charge for the run in Console; the scraper stops cleanly when it is reached and tells you how many results it saved.
Run summary (RUN_SUMMARY)
Every run writes a RUN_SUMMARY record to its key-value store (linked in the run's Output tab): per video the comments and replies saved, pages read, the comment count YouTube shows, and why it stopped (maxCommentsPerVideo, endOfComments, maxResults, chargeLimit, COMMENTS_DISABLED, …); request and failure counts; rows saved, error rows and duplicates skipped; field fill rates; and warnings.
Limits and known gaps
- Rounded like counts: YouTube shows large like counts rounded ("570K"), and
likeCountis that rounded number (570000). Small counts are exact. - Times are estimates: YouTube only shows relative times ("6 years ago").
publishedAtEstimateis computed from that text, so it is only as precise as the unit shown (an "x years ago" comment can be off by months). - Avatar URLs expire:
author.avatarUrllinks to YouTube's image servers and can stop working after some time. Download avatars right away if you need to keep them. - Top order is YouTube's: "Top comments" is YouTube's own ranking and can change between runs. "Newest first" is chronological.
- Very large videos: videos with millions of comments take time to read in full. Use
maxCommentsPerVideoormaxResults. - Public comments only: held-for-review, deleted or members-only content that YouTube doesn't show publicly is not returned.
- Free plan preview: on Apify's free plan a run is a preview: only the first video in your list and at most 20 rows. Any paid Apify plan removes the limit.
Use with AI agents (Apify MCP)
Add the scraper to Claude, Cursor or any MCP client:
https://mcp.apify.com?tools=ragnum/youtube-comments-scraper
Agents can call it with plain inputs (videos, maxCommentsPerVideo, sort) and get typed rows back, for example: "get the 300 newest comments on this video and summarize the main complaints".
FAQ
Why did I get fewer comments than maxCommentsPerVideo? The video has fewer comments than the limit, the run-wide maxResults or your maximum charge was reached, or a request failed (you then get an error row). RUN_SUMMARY.inputs[…].stoppedBy tells you which.
Why is the number of comments lower than the count on the video page? The count on YouTube includes replies and comments YouTube no longer shows. Turn on includeReplies to get replies too.
Do replies count toward maxCommentsPerVideo? No. That limit counts top-level comments only; replies have their own maxRepliesPerComment limit. Both count toward maxResults, and both are charged.
Can I scrape Shorts? Yes. Paste Shorts URLs or IDs like any other video.
Can I get comments from a whole channel or a playlist? Give the video URLs. Channel and playlist links get a free INVALID_INPUT error row.
Can I get comments in another language? Comments are always returned as written. hl only changes YouTube's interface texts such as "3 days ago".
Is the same comment ever charged twice? No. Comments are de-duplicated by commentId within a run, and duplicates are free.
Your responsibilities
This scraper collects only data YouTube shows publicly to logged-out visitors. You are responsible for how you use it: comply with YouTube's Terms of Service and with data-protection law such as the GDPR and CCPA. Comments, names, handles and avatars are personal data of their authors: have a lawful basis before you store or process them, keep only what you need, and honor deletion requests.