The TikTok video scraper API returns per-video metrics for public TikTok videos — exact view, like, comment, share and save counts, plus description, creation time, author with follower count, duration, music track and hashtags — for a list of video URLs or ids.
The word doing the work is exact. TikTok's interface abbreviates everything to "1.2M" and "45.3K"; the underlying payload carries integers, and this collector delivers those. For anything analytical — engagement rate, share-to-view ratio, week-over-week deltas — the difference between 1,200,000 and 1,247,318 is the difference between a chart and a decision. Videos that are private, removed or region-locked come back under failed with TikTok's own reason, never billed as delivered.
$0.004 per delivered video, up to 50 videos per run. Nothing delivered means nothing charged.
You post an array of video URLs or ids, up to 50 per run. Each is fetched as a public page through a residential exit and parsed from its embedded data payload — no login, no cookies, no browser at any point.
The saves field is worth pausing on. TikTok exposes a save (bookmark) count that most scrapers ignore because it is not rendered as prominently as likes, and it is arguably the strongest intent signal on the platform: a save means someone intends to come back. If you are ranking content by genuine resonance rather than by passive scrolling, saves beat likes.
Failures are separated rather than smoothed over. A removed video, a private account or a region lock returns in failed with the platform's reason attached, so a monitoring job can tell "this video was taken down" from "our fetch broke" — which are very different findings.
videos is required: an array of full TikTok video URLs or bare video ids. max_results caps how many are fetched, useful when the array is produced by an upstream job. country sets the exit region, which genuinely changes results here — region-locked videos are visible from some markets and not others.
One row per video. views, likes, comments, shares and saves are exact integers from the payload rather than parsed display strings. author_followers is stamped on every row, which means engagement rate is computable from a single row without a second lookup. hashtags comes back as a list rather than as raw text, music and music_author identify the audio — the field trend research runs on, since a sound is usually the unit of virality rather than a creator. created_at lets you compute velocity properly instead of comparing a day-old video with a year-old one.
$0.004 per delivered video ($4 per 1,000). Nothing delivered means nothing charged, and the $2 monthly free credit covers roughly 500 videos here. Volume tiers take up to 30% off.
$0.004 per delivered video, $4 per 1,000. Tracking 250 videos daily across a campaign is $1 a day. A one-off analysis of 5,000 videos in a niche is $20.
Failed videos — removed, private, region-locked — are never billed, so an ageing URL list costs you coverage rather than money. The free $2 monthly allowance covers about 500 videos.
TikTok's Display API returns video data for users who have authorised your application, which makes it the right tool for a creator dashboard where people connect their own accounts, and useless for analysing videos you have no relationship with. The Research API does expose public video data with real analytical depth — and is restricted to approved non-profit academic institutions in eligible regions.
Marketing teams sit in the gap between those two. They need engagement metrics on videos in their category, on competitors' content, on creators they are considering paying — none of which involves anyone authorising anything. That gap is what this collector fills, using only what a public visitor can see.
The open-source TikTok video scrapers have a consistent lifecycle: impressive demo, six weeks of service, silent failure. The causes repeat:
Reading the public page payload from a residential exit sidesteps the signing problem entirely — which is why this collector is scoped to public videos and says so.
Track the videos in a paid campaign daily. views against author_followers gives reach beyond the creator's own audience, which is what you actually bought.
Rank a competitor's catalogue by engagement rate rather than by view count to find what genuinely resonated rather than what got pushed.
Aggregate music and hashtags across a sample to spot a trend while it is still climbing — sounds usually move before creators do.
Before paying a creator, sample twenty of their videos. Views wildly inconsistent with follower count is the pattern worth catching.
Up to 50 videos per run. Bigger lists are more runs, parallel up to your plan's rate limit — 60 requests/minute on pay-as-you-go, up to 1,200 on the top tier. Only public videos are in scope; anything private, removed or region-locked is returned in failed with TikTok's reason and is not billed.
Legally: metrics on public videos are public data, and this returns metadata rather than media. But the author fields identify people, which brings data-protection law into play, and TikTok's terms restrict automated collection. Video content itself is copyrighted by its creator — nothing here downloads it. Not legal advice.
There is a free allowance of $2 a month with no card, about 500 delivered videos at $0.004 each. The free GitHub projects mostly depend on bundled browsers or reimplemented request signers, which is why they tend to stop working after a TikTok client update.
Yes. They come from the video page's data payload as integers, not from the abbreviated display strings the interface renders. That distinction matters for any ratio or delta you plan to compute.
TikTok's bookmark count. It is under-used because the interface does not emphasise it, and it is arguably the strongest intent signal on the platform — a save means someone means to come back to it. Ranking by saves surfaces different content than ranking by likes.
They come back under failed with TikTok's own reason and are not billed. That separation is deliberate: a monitoring job needs to distinguish 'this was taken down' from 'our fetch failed', because they mean different things.
$0.004 per delivered video — $4 per 1,000 — with no subscription, and failures are never charged. Volume tiers take up to 30% off, and GET /v1/scraper/collectors returns the price your key actually pays.
No. It returns metrics and metadata — counts, description, author, music, hashtags, duration and the cover image URL. The media itself is the creator's copyrighted work and is not fetched.
No. Your QuantumProxies key is the only credential, and no TikTok session is involved. The official Display API needs users to authorise your app, and the Research API is limited to approved academic institutions — neither covers reading arbitrary public videos.
Skip the marketing. QuantumProxies.io publishes a machine-readable site map, Markdown for every page, a free MCP server, and APIs billed from the same balance as your proxies.
Connect in one command: npx -y quantumproxies-mcp