Are replies included?

Yes. Replies come back as their own records with type set to reply and reply_to_id pointing at the parent comment.

Can I get only recent comments?

Set newer_than to a date or a period like 30 days. The comments come back newest first.

Do I pay for lookups that come back empty?

No. You are charged per result, so a request that finds nothing costs nothing. Testing this against a messy list is free to get wrong.

2 points per returned comment, which is $2 per 1,000, capped by max_comments.

Social media

YouTube Comments API and MCP — Comments and Replies on a Video

Send a YouTube video URL and this source returns its comments and replies as data: author handle, text, likes, reply count, whether the creator hearted it, whether the author is the channel owner, and when it was posted.

Run it in the app Read the API docs

You send: A YouTube video URL
You get back: Comments and replies with author, text, likes and reply count
Coverage: Comments on public YouTube videos with comments switched on.
Price: 2 points per returned comment, up to 1,000 per request — $2 per 1,000; Billed per comment actually returned, capped by max_comments. A video with comments turned off costs nothing.

Comparison

Ways to get this data

Three approaches teams use today, and where each one runs into trouble.

Approach	What it takes	Cost	Trade-off
Reading comments by hand	Scroll and copy	Free, and slow	Not workable past a few dozen comments
The YouTube Data API	A Google Cloud project, API key and quota	Free within a daily quota	Paginated and quota-limited across many videos
NeuralVerge	One request per video	$2 per 1,000, misses not billed	Publish times come back relative ("9 months ago"), not as exact timestamps

Response

What you get back

Every field in the example response below.

Field	Type	Description
`sort_by`	string	The order used: top comments, or newest first.
`total_items`	number	How many comments came back.
`items[].id`	string	YouTube's comment ID.
`items[].type`	string	comment or reply.
`items[].reply_to_id`	string \| null	The parent comment, for replies.
`items[].author`	string	The commenter's handle.
`items[].text`	string	Comment text.
`items[].published_time_text`	string	When it was posted, as YouTube shows it (e.g. 9 months ago).
`items[].vote_count`	number	Likes on the comment.
`items[].reply_count`	number	Replies to the comment.
`items[].has_creator_heart`	boolean	Whether the creator hearted the comment.
`items[].author_is_channel_owner`	boolean	Whether the channel owner wrote it.
`items[].video_title`	string	The video the comment is on.
`items[].video_comments_count`	number	Total comments on the video.

Response

Example response

A real response, as returned.

{
  "session_id": "e72bce7d-3a66-4a78-b50a-88d1bcbb91fe",
  "kind": "youtube_comments",
  "url": "https://www.youtube.com/watch?v=RIRyuq-AKOU",
  "max_comments": 2,
  "sort_by": "TOP_COMMENTS",
  "partial": false,
  "machine": {
    "total_items": 2,
    "items": [
      {
        "id": "Ugz_0evfWBOA6L1RlKR4AaABAg",
        "type": "comment",
        "reply_to_id": null,
        "author": "@dylan…",
        "text": "The AI avatar made me lose trust i gotta be honest. Just letting the account owner know. Im sure the info is great and in-depth but it begs was this made entirely with AI and not actually opinions and experience developed by a human.",
        "published_time_text": "9 months ago",
        "vote_count": 62,
        "reply_count": 5,
        "has_creator_heart": false,
        "author_is_channel_owner": false,
        "video_id": "RIRyuq-AKOU",
        "video_title": "Best CRM Software for Small Business for 2026 (Top 5 Tested & Compared)",
        "video_comments_count": 39
      },
      {
        "id": "UgwXgsbBvTJvsJo51kR4AaABAg",
        "type": "comment",
        "reply_to_id": null,
        "author": "@beth…",
        "text": "Felt like a review video at first, then became an obvious GHL advert.",
        "published_time_text": "6 months ago",
        "vote_count": 27,
        "reply_count": 1,
        "has_creator_heart": false,
        "author_is_channel_owner": false,
        "video_id": "RIRyuq-AKOU",
        "video_title": "Best CRM Software for Small Business for 2026 (Top 5 Tested & Compared)",
        "video_comments_count": 39
      }
    ],
    "errors": []
  },
  "total_points": 4
}

Integration

How to run it

The same lookup works from the app, the API, or as a tool an agent can call mid-task.

This endpoint is synchronous: the call blocks until the lookup completes and returns the records directly. If a lookup hits its time budget before finishing, the response comes back with partial: true and whatever it collected so far — re-run with smaller limits for a complete result. url is required. Optional: max_comments (1-1,000, default 20), sort_by (TOP_COMMENTS; leave it out for newest first), newer_than (a date or a period like 30 days — returns newest first).

curl -X POST https://api.neuralverge.ai/functions/v1/run-youtube-comments \
  -H "Authorization: Bearer YOUR_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{ "url": "https://www.youtube.com/watch?v=RIRyuq-AKOU", "max_comments": 100, "sort_by": "TOP_COMMENTS" }'

Use cases

What teams use it for

Where this field shows up in an actual workflow.

Product review sentiment

Read what viewers say under reviews of your product or a competitor's.

Objection mining

Collect the doubts and questions viewers raise about a category.

Content feedback

See how an audience reacted to a launch video or webinar.

Creator due diligence

Check whether a creator's audience trusts them before a sponsorship.

Coverage & limits

Coverage, freshness and limits

What this source covers, how fresh it is, and where it stops.

Coverage

Comments on public YouTube videos with comments switched on.

Freshness

Read at request time.

Handling

This source returns publicly posted comments, which are written by private individuals. Use them for legitimate business purposes under your own lawful basis, honour deletion and opt-out requests, and check the acceptable use policy. Every record is stamped with the video it was read from and a session id you can trace.

What it will not do

—Publish times are relative text ("6 months ago"), not exact timestamps.
—Videos with comments turned off return nothing — and nothing is billed.
—Authors are handles only; no profile details come back.

FAQ

Questions, answered

Yes — every Data Source endpoint, including this one, is synchronous: the lookup runs inside the request and the result comes back in the same response, no polling. If a lookup hits its time budget before finishing, the response comes back with partial: true and whatever records it collected so far — re-run with smaller limits to get a complete result instead of a partial one.