TikTok Video Scraper | Post Stats and Downloads

TikTok Video Scraper reads the public video and slideshow URLs you supply and returns captions, engagement counts, author and music details, subtitles and optional media files. Every input position gets one item, in input order, including missing posts and duplicates; those status items and duplicates have no result charge.

Try a TikTok URL

Run the prefilled NBA post to inspect the returned fields. The form prefill is an example only; an API request with {} or postUrls: [] reads nothing and exits successfully with guidance.

{
  "postUrls": ["https://www.tiktok.com/@nba/video/7693329153680035102"],
  "proxyConfiguration": {"useApifyProxy": true}
}

Input

Field	Type	Default	What it is
postUrls	array	—	Standard /@author/video/id and /@author/photo/id URLs. Prefilled example, no API default. Input order and duplicate positions are preserved.
includeSubtitleText	boolean	false	Retrieve TikTok-provided WebVTT subtitle text where present; no audio transcription.
downloadVideos	boolean	false	Store complete video files during the primary reading session.
downloadCovers	boolean	false	Store cover images during the primary reading session.
downloadSlideshowImages	boolean	false	Store each slideshow image during the primary reading session.
videoKvStoreIdOrName	string	—	Download store name (names only); omitted uses the run's default key-value store.
proxyConfiguration	object	—	Optional buyer proxy settings. Omitted uses Apify's automatic pool, without a fixed proxy group. The form prefill enables Apify Proxy.

Output

Each readable post has inputUrl, one-based inputPosition, UTC scrapedAt, string id, source-author url, and postType (video or slideshow). The requested ID is checked against the source ID. A wrong URL handle does not replace the source author.

Example excerpt showing field types, not a measured post:

{
  "inputUrl": "https://www.tiktok.com/@example/video/1234567890123456789",
  "inputPosition": 1,
  "scrapedAt": "2026-10-12T00:00:00Z",
  "id": "1234567890123456789",
  "postType": "video",
  "url": "https://www.tiktok.com/@example/video/1234567890123456789",
  "text": "An example caption",
  "playCount": 1200,
  "collectCount": 0
}

Fields	Meaning and units
text, textLanguage, locationCreated, isAd	Caption, source language, source country code and source ad flag. An absent boolean stays null or absent.
createTime, createTimeISO	Unix seconds and ISO UTC time of creation.
playCount, diggCount, commentCount, shareCount, collectCount, repostCount	Integer engagement counts; zero is valid.
author	ID, unique handle, nickname, secUid, verified flag, signature, private-account flag, creation time (Unix seconds), seller/room fields, avatars and profile URL.
authorStats	Integer followers, following, hearts, videos, diggs and friends. These statistics arrive rounded from TikTok.
music	String ID, title, author, original flag, audio URL, cover URL and duration in seconds.
video	Duration in seconds, dimensions in pixels, definition, format, bitrate in bits per second, cover and original cover. Slideshows have source duration zero.
hashtags, mentions	Hashtag ID/name/title/cover and mentioned handle/ID, where supplied.
slideshowImages	Ordered URL alternatives for each slideshow image.
subtitles	Language, source and URL; optional WebVTT text or status: not_delivered.
media	Requested file outcomes. A delivered file has kind, store ID, key, retrievable URL and size in bytes. Files not delivered have an explicit status. Store access requires your Apify permissions.

Unreadable inputs carry inputUrl, inputPosition and a status: missing, private, gated, failed or unsupported_url. A duplicate repeats the first read's fields, with status: duplicate_of and duplicateOf naming its first input position. It causes no additional source request or charge. After a charge limit binds, remaining positions become unbilled failed items. Already admitted reads settle, and no new source request starts. Additional reasons appear in ERRORS; run counts and bytes appear in SUMMARY, including primaryReads (post-page requests, excluding files) and softRefusals (HTTP 200 pages under 2,000 bytes without the post-hydration marker). An empty-input run records zero reads and refusals. A dataset or charging fault ends the run FAILED with a fixed message; earlier rows remain available. An optional file-store failure retains the metadata and reports the file as not delivered.

Billing and files

Readable, distinct posts use the result event. Status items and duplicate positions have no result charge. A separate Actor start event applies to every run, including runs with no readable post. A complete stored video, cover or slideshow image uses its corresponding download event only after storage succeeds and the row is delivered. Files whose result-plus-download cost does not fit the remaining charge allowance are not retrieved or stored. Stored keys include a content digest so later downloads do not replace earlier different files. Download sizes are reported in bytes. Subtitle text is included when selected and available. Browser-fallback reads are not billed as extra results. Prices are set in the Actor's Pricing tab.

Public access and limits

The primary uses anonymous browserless HTTP through Apify's automatic proxy pool, with at most two parallel sessions, each paced at least 1.1 seconds after its previous request completes. This pacing is deliberate, and it makes larger batches slower than the other TikTok video Actor we measured on the same URLs. In our tests, run start to complete dataset took 14-61 seconds for 10 URLs (3 runs; the other Actor 23.9 seconds at p50 and p95) and 55-75 seconds for 50 URLs (4 runs; the other Actor 33.3 seconds at p50 and 62.7 seconds at p95). These small samples are not performance guarantees. Rows are delivered in input order. Direct connections and buyer-supplied custom proxy URLs stay sequential to pace a shared egress identity.

An HTTP block, non-login redirect, maintenance shell or missing hydration sends the input to our shared browser service once, at no extra result charge. Source-reported missing, private and login outcomes do not trigger a browser retry. The service uses its own datacenter pool, not the buyer's proxy configuration. A blocked primary session is replaced for the next distinct input. A redirect to /notfound alone does not decide that a post is missing; the browser's hydrated answer does. Redirect diagnostics keep the Location path and login classification in ERRORS, without the query.

Some anonymous reads return an HTTP 200 page under 2 KB without post data. In our October 12-13, 2026 release tests, 9 of 255 post-page reads took this path (3.5%; 95% interval 1.9-6.6%), and the browser fallback served each of them. This dated sample is not a reliability guarantee; each run's SUMMARY reports its own primaryReads and softRefusals. Browser-served posts cannot include downloads (see below).

The shared browser service cannot keep page cookies across DOM and file operations. A browser-served post keeps its metadata, but requested video, cover and slideshow-image downloads report media.status: session_unavailable and carry no download charge. Cookie-free subtitle reads still work when TikTok supplies them. Page cookies stay in memory within the primary run and never appear in the dataset, stores or logs. TikTok media and subtitle URLs expire; the stored file, not the source URL, is the deliverable.

Short/share links, profile feeds, comments, search, AI transcription and AI description/summary are not supported. Related videos are not provided; their request needs a browser-side signer the Actor does not run. Private, sensitive and age-gated post classes have not been measured with live fixtures.

The measured pages lacked isPinned, isSponsored, isStory, bioLink, hasActiveStory, commerce and shortDramaSeriesInfo fields, and this release does not provide them. The Actor does not derive them from seller, story-status or anchor values. Do not read their absence as false.

TikTok's US Terms (last updated July 15, 2026) prohibit automated extraction without written approval and reverse engineering without written authorization. Its robots file disallows named crawlers including ClaudeBot and GPTBot; the wildcard section does not disallow the video-page path, and that wildcard rule is not written approval under the Terms. The Actor does not log in.

Help

Check the input position's status, ERRORS and SUMMARY. For support, include the run URL and the failing input position through the Actor's Issues tab.