KonbiniAPI

The TikTok API for products and pipelines.

Retrieve supported public TikTok data — profiles, videos, comments, transcripts, hashtags, audio tracks, and search — through REST endpoints that return one consistent response model.

  • 22 supported endpoints
  • Public data only
  • Fetched at request time
GET/v1/tiktok/videos/{videoId}200OK
{
  "type": "Video",
  "content": "Drop your questions below! 🚀",
  "published": "2026-07-27T16:25:49.000Z",
  "duration": 13,
  "viewCount": 861100,
  "likeCount": 73600,
  "commentCount": 8318,
  "shareCount": 1417,
  "attributedTo": {
    "preferredUsername": "nasa",
    "isVerified": true
  }
}

What you can pull from TikTok.

More is reachable on TikTok than on any other platform here. A profile lookup returns bio, follower and like counts, verification status, and media counts. Video feeds paginate by cursor and sort by newest, popular, or oldest. Comment threads and their replies are separate endpoints, so a thread can be walked to whatever depth a workload needs. Beyond a creator’s own content there are hashtag and audio endpoints, which is how discovery work usually starts: find the sound or tag, then read the videos using it.

Video objects carry the media files in several qualities, the audio track with its own author, hashtags as first-class tags, and the engagement counts. Transcripts are a separate call and return WebVTT — the auto-generated one under the original language, or a machine translation under a BCP47 language code.

Every response uses the same ActivityStreams 2.0-based model as the other supported platforms, so a TikTok integration reuses the parsing you already wrote. Fields specific to TikTok stay as documented first-class fields rather than being flattened away.

TikTok data
  • Profiles, follower and following lists
  • Video feeds, single videos, and download URLs
  • Comments and threaded replies
  • Transcripts in WebVTT
  • Hashtag, audio, and keyword search
  • Stories, collections, and live status

All 22 TikTok endpoints.

Grouped by what they return. Every one has its parameter contract, a request you can copy, and an example response.

Profiles

3

The creator record itself, plus the follower and following lists that hang off it.

Feeds

6

Cursor-paginated views of what an account has posted, liked, reposted, collected, or is streaming right now.

Posts

4

One video by id, and everything attached to it: the download URL, the WebVTT transcript, and the videos inside a collection.

Comments

2

Top-level comments on a video, and the reply chain under any one of them.

Discovery

7

Where a run starts when you have no id yet — hashtags, audio tracks, and keyword search over videos, users, and mixed content.

How the calls fit together.

Few TikTok jobs are a single request. Most are two or three chained, each call keyed by an id the one before it returned.

Vet a creator

Size the account, read what it actually publishes, then read how its audience responds.

  1. 1Get user profile/v1/tiktok/users/{username}
  2. 2Get user videos/v1/tiktok/users/{username}/videos
  3. 3Get video comments/v1/tiktok/videos/{videoId}/comments

Follow a sound

Read the track for its title, artist, and how many videos use it, then page those videos to see how far it has spread.

  1. 1Get audio details/v1/tiktok/audios/{audioId}
  2. 2Get videos with audio/v1/tiktok/audios/{audioId}/videos

Turn a video into text

Find candidates by keyword, then take the transcript straight off the ids the search already returned.

  1. 1Search videos/v1/tiktok/search/videos
  2. 2Get video transcript/v1/tiktok/videos/{videoId}/transcripts/{language}

What shapes your TikTok integration.

Page sizes, ordering, and what the platform itself withholds — the details that decide how you build, settled before you write a line of code.

01Page sizes differ by endpoint
Video feeds and collections return at most 35 items per page, follower and following lists 30, comments and replies 50, and audio video lists 30. User search has a fixed page size of 10, which is the platform’s own limit.
02Likes and reposts can be hidden by their owner
A user can set their likes or reposts to private. Those endpoints then return an empty list rather than an error, so treat an empty page as “not shared” rather than as “none”.
03Stories and live status are transient
Stories expire after 24 hours, and the live endpoint returns 404 when the account is not currently streaming. Both are point-in-time reads, not history.
04Download URLs are short-lived
The video download endpoint returns a URL you can fetch with a plain GET, with no cookies or headers of your own. It is valid for a few hours, so resolve it at the point of download rather than storing it.

Build on TikTok data.

Get a free API key and call any of these endpoints in minutes. No credit card required.