TikTok Research API alternative: what to use when you don't qualify

The TikTok Research API is the most complete official source of public TikTok data: date-bounded video queries, paged follower lists, comments, liked and reposted videos. It is also closed to many of the people who look for it. Agencies, startups, brands and researchers outside the eligible regions or institutions can't apply, and nobody may use it commercially.

If you qualify, use it: TikTok announced it free of charge for researchers and publishes no price. This page is for everyone else, and for researchers who want to know what a commercial API can and cannot replace.

Who can use the Research API

• Where: researchers in the US, the EEA, the UK, Canada, Switzerland, Norway, Iceland, Liechtenstein and Brazil. In Brazil, the research must be about youth safety.
• Who: academic institutions in the US, EEA, UK, Canada or Switzerland, and not-for-profit or independent research institutions in the EU. Ph.D. candidates need an advisor's endorsement letter.
• For what: non-commercial research only. Applicants must be independent of commercial interests; creators, advertisers and commercial users are not eligible.
• How: create a TikTok for Developers account, submit an application with a research proposal, and wait. TikTok says to expect an answer within about 4 weeks. An approved researcher can add up to 9 collaborators, who share the team's quota.

The short version

ScrapingBot vs the TikTok Research API at a glance

Endpoint by endpoint

Every ScrapingBot TikTok call is a POST to https://scrapingbot.io/api/v1/tiktok naming an endpoint, for 1 credit. This is how the Research API's endpoints map onto them:

Video fields map closely. Counts come as TikTok displays them, so large ones are rounded (1.8M arrives as 1800000):

One question, both ways: a keyword's videos from the last 30 days

With the Research API, "videos about outdoor cooking from the last 30 days" is one request to its video query endpoint, bounded by dates. In the format its reference documents, the body is:

{
  "query": {
    "and": [{ "operation": "EQ", "field_name": "keyword", "field_values": ["outdoor cooking"] }]
  },
  "start_date": "20260928",
  "end_date": "20261027",
  "max_count": 100
}

It goes out with a client access token, and the fields you want are listed in the URL, such as fields=id,video_description,create_time,username,view_count,hashtag_names. Each request returns up to 100 videos and a cursor for the next.

ScrapingBot has no date filter. You page through TikTok's own search, which ranks by relevance across all dates, and keep the recent videos:

from datetime import datetime, timedelta, timezone

import requests

API = "https://scrapingbot.io/api/v1/tiktok"
HEADERS = {"x-api-key": "YOUR_API_KEY"}


def search(keywords, pages=4, region="US"):
    """Videos TikTok's search shows for a keyword, de-duplicated: 1 credit per page."""
    videos, params = {}, {"keywords": keywords, "count": 30, "region": region}
    for _ in range(pages):
        body = requests.post(API, headers=HEADERS, timeout=60,
                             json={"endpoint": "/feed/search", "params": params}).json()
        if not body.get("success"):
            raise RuntimeError(body.get("error"))
        data = body["data"]
        for v in data["videos"]:
            videos.setdefault(v["video_id"], v)
        if not data.get("hasMore"):
            break
        params = {**params, "offset": data["cursor"], "search_id": data["searchId"]}
    return list(videos.values())


since = datetime.now(timezone.utc) - timedelta(days=30)
found = search("outdoor cooking")
recent = [v for v in found if datetime.fromtimestamp(v["create_time"], timezone.utc) >= since]
print(f"{len(found)} videos, {len(recent)} posted in the last 30 days")
for v in sorted(recent, key=lambda v: v["play_count"], reverse=True)[:5]:
    tags = sorted({t["hashtag_name"] for t in v.get("text_extra") or [] if t.get("hashtag_name")})
    print(f"{v['play_count']:>10,}  @{v['author']['unique_id']:<22} {' '.join('#' + t for t in tags[:3])}")

176 videos, 44 posted in the last 30 days
62,500,000  @outruncook             #citytastetest #cooking #easyrecipetiktokcontest
 9,200,000  @mountainchef420        #firecooking #mountainchef #outdoorcooking
 9,000,000  @outruncook             #citytastetest #cooking #easyrecipetiktokcontest
 4,300,000  @outdoor_chef7          #cooking #cookingforest #outdoorcooking
 3,500,000  @firerun.crew           #campfirecooking #campingfood #cookinginnature

That run cost 4 credits, well under a cent on Starter. The four pages came back with 30, 96, 95 and 93 videos (TikTok often returns more than count), 176 of them unique. Only 44 were from the last 30 days and 17 from the last week; the oldest went back to 2022. Each video arrived with its play URL, cover and sound.

The difference matters for research. TikTok's search is what a user sees: ranked by relevance and partial. The Research API's query is a filter over TikTok's index, so it is the right tool for a defensible sample ("every video with this hashtag posted in the UK in May"). For a product, a dashboard or a monitor that wants the popular videos a user would find, along with their media, search is the closer match.

Followers and following

This is the largest gap. The Research API pages through an account's followers 100 at a time, with 2 million records a day to spend. ScrapingBot's /user/followers and /user/following return one page of up to 50 accounts. Asking NASA's followers for 50 returned 50, with total at 2,141,592; there is no cursor to go further.

Each of those accounts carries more than the Research API's username and display name: its own follower and following counts, total likes (total_favorited), video count (aweme_count), bio, region and verified badge. For "who are some of this account's followers, and how big are they", one call is enough. For a follower graph beyond that first page, you need the Research API. (On Instagram, ScrapingBot does page through whole lists: see exporting Instagram followers.)

What it costs

• Research API: no fee for approved projects (TikTok announced it free of charge in 2023 and lists no price), within 1,000 requests a day (100,000 records) plus the follower-list quota.
• ScrapingBot: 1 credit per TikTok call, whatever it returns. A search page of 30 to 96 videos, a profile, a page of 50 comments or 50 followers each cost 1 credit. Starter is $49.99 for 275,000 credits, so 1,000 calls is about 18 cents. Failed calls are refunded, and the 100 free credits are enough to try every endpoint.

The rules that come with Research API data

Approved researchers accept TikTok's Research Tools terms. Among them: refresh the data at least every 30 days and delete what is no longer available, keep it no longer than the research needs, send TikTok a copy of any publication at least 7 days before it appears, and don't combine the data with other information about individual users. Budget for the refresh work if you plan a long study. ScrapingBot data comes with ScrapingBot's terms, and your own obligations under privacy law still apply to what you store.

Where to go from here

• Try the search above in the playground, or follow the full guide to searching TikTok videos by keyword with Python.
• Track one sound instead of a keyword: every video that uses a TikTok sound, or the free TikTok sound tracker.
• Need engagement numbers for a creator? The free TikTok engagement rate calculator, or follower and like counts with Python.
• Comparing commercial TikTok APIs? The best TikTok data APIs in 2026, and the TikTok API page for every endpoint.

Common questions