Why the same TikTok data costs 100x more from one API than another

Looking up one TikTok profile costs about $0.00002 through one API and $0.002 through another: a hundredfold difference for the same public data. Most of the gap isn't about data quality. It comes from how the price is quoted: a subscription spread over a quota, how many records one call returns, how many companies sit between you and the scraper, whether you pay for automatic failover, and whether the route is the platform's own licensed API.

Our database has 51 priced routes to a TikTok profile alone, and 1,460 TikTok endpoints from 106 providers. Here is what that spread looks like and why it exists.

The spread, in one table

One job: get one TikTok account's profile. Price per call at each seller's entry price, from /api/cheapest.

The RapidAPI rows are the plan price divided by the plan's quota. That is only true if you use the whole quota, which is the first reason on the list.

1. Subscription quotas are amortised per call

A RapidAPI listing sells a monthly plan, not a call. tiktok-scraper7's PRO plan is $59 for 3,000,000 requests. If you make all 3,000,000 calls, each costs $0.0000197. If you make 10,000, each costs $0.0059, which is 300 times more and dearer than every pay-per-call seller in the table except Socialfetch.

So the cheapest-looking rows are cheapest for heavy, steady users. The working is in Subscription quota vs pay per call: the break-even math.

2. One call can return 1 record or 30

Price per call says nothing about what comes back. We made two real TikTok calls through our gateway for this post:

• tiktok-scraper7 feed/search for "coffee" returned 19 videos for $0.0000197: about $0.001 per 1,000 videos.
• tiktok-api6 user/videos for @mrbeast returned 30 videos for $0.001998: about $0.067 per 1,000 videos.

A profile lookup returns one record whatever you pay, so per-call and per-record prices match. For lists, the number that matters is dollars per 1,000 records. We worked it through for six real calls in What one API request really costs.

3. Reseller layers add margin

A lot of data APIs resell another company's API under their own name. Each layer adds its own margin, billing and support. Our database names the source where the seller states it:

• Pipeworx sells a TikTok profile tool built on ScrapeCreators for $0.005 a call. ScrapeCreators' own price is $0.00188, so the extra layer costs about 2.7x.
• TaskFuel lists TikTok profile videos "via Scrape Creators" at $0.02 a call, about 10.6x the source price.
• Going the other way, some marketplace sellers also sell direct: our research records that the operator of twitter-api45 sells the same API at scraper.tech, and the operator of tiktok-api23 sells direct as Tikfly.

A reseller isn't automatically a bad deal: one key, one invoice and one schema across many platforms are worth something. But when two prices differ tenfold for the same rows, check what sits underneath.

4. Failover costs extra, and it shows in the price

Every scraper fails some of the time. Some sellers retry on a second provider automatically and charge that provider's price when they do. AnyAPI is open about it: each endpoint lists its lanes (upstream providers), each lane's price and its 30-day uptime. For TikTok keyword search its catalogue shows:

The headline price is the cheapest lane; a call that fails over is billed at the lane that serves it, up to $0.0036 for this endpoint. Part of what you pay for is a higher success rate, and failed calls are where the real cost hides. More on that in The hidden costs of a data API.

5. Official API pricing sets the ceiling

Our database has no priced official route for TikTok data: the /api/cheapest answer for tiktok.com returns "official": null. X has one, and it shows how far apart the two worlds are. X's pay-per-use API charges $0.005 per post read and $0.010 per user. Through our gateway, a twitter-api45 search for "web scraping api" returned 19 posts for $0.00009. That works out to about $0.0047 per 1,000 posts, against $5 per 1,000 from X.

The official price pays for a licence, published terms and an API that doesn't break when the site changes its front end. Unofficial APIs read public pages and don't pay for those things. If your use needs the licence, the official price is the real one, so decide that before you compare.

What to do with this

• Compare dollars per 1,000 records, not per call. Run one real call and count what comes back.
• Price at your own volume. A quota price assumes you use the whole quota.
• Ask what is underneath. If two sellers return the same fields, one may be reselling the other.
• Count failures. The cheapest lane is often the least reliable one.

How FetchAPI prices these calls

We hold the entry plans and meter each successful call at the plan's per-call price (for example $0.0000197 for a tiktok-scraper7 call) times our markup, which is currently 1.0. Failed calls are free, and every key has a monthly spending cap. That gives you the full-quota per-call price without buying the subscription. The prices for TikTok are on /apis/tiktok and the tiktok-scraper7 provider page. Here is the call we made:

curl -s "https://fetchapi.co/api/gw/rapidapi-tiktok-scraper7/feed/search?keywords=coffee&count=20&cursor=0&region=us" \
  -H "x-api-key: $FETCHAPI_KEY"

The answer's meta.cost_usd is the metered price of that call. Caveats: RapidAPI per-call prices assume the plan's quota is used, several rows above are estimates from plan pages (dates in the sources below), and our markup can change. The charge recorded with each call is what you pay. Browse every endpoint in the catalog, plug it into an agent with MCP, or get a free key.