How to Get TikTok Video and Comment Data Through an API

To get TikTok video and comment data programmatically, use Scavio's TikTok API. POST /api/v1/tiktok/video with a video id returns the video's description, author, sound and statistics (views, likes, saves, shares and comment count); POST /api/v1/tiktok/video/comments returns up to 50 comments a call with text, likes, reply counts and timestamps, and /video/comments/replies opens any thread. Hashtag and search endpoints find the videos, and /video/transcript returns what is said in them. Every call is 1 credit, about $0.002 on the Growth plan billed yearly.

TikTok's own APIs do not cover this for most teams: the Research API is for approved non-commercial researchers, and the Display API reads only the videos of a user who logs in to your app, with no comments at all. Rejected for TikTok's Research API? covers who qualifies.

Step 1: Find videos by hashtag or keyword

Resolve the hashtag name to its id first:

curl -s https://api.scavio.dev/api/v1/tiktok/hashtag \
  -H "Authorization: Bearer $SCAVIO_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"hashtag_name": "booktok"}'

On October 27, 2026, data.challengeInfo.challenge.id came back as 1622962893630470, and statsV2 reported 88,078,371 videos and 562,242,547,824 views. Pass the id to hashtag videos:

curl -s https://api.scavio.dev/api/v1/tiktok/hashtag/videos \
  -H "Authorization: Bearer $SCAVIO_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"hashtag_id": "1622962893630470"}'

Each page returned 20 videos in data.aweme_list, with cursor and has_more for the next page. Three pages gave 60 rows and 57 distinct videos, so key your store on aweme_id. The feed is ranked by TikTok, not sorted by date: those 57 ran from August 2021 to the day we pulled them. For a keyword instead of a hashtag, send {"keyword": "book recommendations"} to /api/v1/tiktok/search/videos; it returned 12 videos with the same statistics, each under search_item_list[].aweme_info.

Step 2: Read one video's statistics

curl -s https://api.scavio.dev/api/v1/tiktok/video \
  -H "Authorization: Bearer $SCAVIO_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"video_id": "7456818947573222689"}'

The response sits under data.aweme_detail. Trimmed to the fields most pipelines keep, as captured:

{
  "aweme_id": "7456818947573222689",
  "create_time": 1736175971,
  "region": "GB",
  "author": { "unique_id": "fionalucasauthor" },
  "music": { "title": "original sound - fionalucasauthor", "id_str": "7456819020759730977" },
  "statistics": {
    "play_count": 1269637,
    "digg_count": 47880,
    "comment_count": 4445,
    "share_count": 2126,
    "collect_count": 6998,
    "download_count": 371,
    "whatsapp_share_count": 285
  }
}

digg_count is likes and collect_count is saves. Across the 57 #BookTok videos the median video drew 91.5 likes, 7.1 saves, 6.0 shares and 0.71 comments per 1,000 views, so saves and shares carry more signal than comment counts; the chart is in the best TikTok video and comment APIs. video.duration is in milliseconds (110300 here).

Step 3: Pull the comments

curl -s https://api.scavio.dev/api/v1/tiktok/video/comments \
  -H "Authorization: Bearer $SCAVIO_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"video_id": "7456818947573222689", "count": 50}'

count takes 1 to 50 and defaults to 20; a 50-row page took 0.8 s server time. The first page also reports total (4,461 here). One comment, trimmed (commenter fields omitted):

{
  "cid": "7462505518641660715",
  "text": "Helpful tips! Thanks!",
  "digg_count": 94,
  "reply_comment_total": 0,
  "create_time": 1737499986,
  "comment_language": "en",
  "is_author_digged": true,
  "is_high_purchase_intent": false
}

Send the returned cursor back for the next page and stop when has_more is 0. The cursor is an offset into a ranked list that keeps moving as people like and reply, so deduplicate on cid. is_author_digged tells you the creator liked the comment, and is_high_purchase_intent is TikTok's own flag for comments that read like buying interest. Our post on why a TikTok comment pull returns about half the count measures how much of total is replies.

Step 4: Open the replies

curl -s https://api.scavio.dev/api/v1/tiktok/video/comments/replies \
  -H "Authorization: Bearer $SCAVIO_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"video_id": "7456818947573222689", "comment_id": "7457546236569355041"}'

Replies are not inlined in the comments payload. The comment id above belongs to the video's top comment (162 likes), which reported 11 replies; this returned the 10 still visible, in the same comment shape, including one from the creator. Call it only where reply_comment_total is above zero.

Step 5: Get what the video says

curl -s https://api.scavio.dev/api/v1/tiktok/video/transcript \
  -H "Authorization: Bearer $SCAVIO_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"video_id": "7456818947573222689"}'

The 110-second video returned available: true, language_code: "en", is_auto_generated: false, 38 timestamped segments and a 340-word text. The first segment, as captured:

{ "start": 0.08, "end": 3.92, "text": "so if you're new to BookTok and you want to know how do you join BookTok" }

Pair the transcript with the comments and you have both sides of the conversation for an LLM to summarise.

All the endpoints for this job

Every TikTok endpoint costs 1 credit; the full list of 24 is on the TikTok API page.

Getting started

1. Sign up at dashboard.scavio.dev/sign-up: 50 free credits, no card required.
2. Copy your API key from the dashboard.
3. Send a POST to /api/v1/tiktok/video/comments with a video id and your key in the Authorization header.

Try the TikTok API free in the playground first, or call it from Claude and other agents through the MCP server.

Reference: TikTok API overview · video · comments · replies · transcript · hashtag videos