TikTok LIVE Discovery & Monitoring Scraper

Discover trending TikTok LIVE rooms, check whether creators are live, and search LIVE by keyword — with viewer counts, creator profiles, playback URLs, and an explainable priority score on every room.

TL;DR

• Three ways to find rooms — trending (no keyword needed), a watchlist of creators, or keyword search. Combine them in one run.
• No mode to select — fill the fields you care about and the Actor runs the matching discovery.
• One flat row per live room, 51 fields, ready for Sheets, Excel, BI tools, or an LLM prompt.
• Momentum, not just size — every room reports how long it has been live and how fast it is pulling viewers.
• Playback URLs included — HLS, FLV and RTMP, with quality, resolution and an explicit expiry timestamp.
• Every room is scored 0–100 with a plain-English reason, so you know which ones are worth opening.
• No TikTok login, cookies, or API key required.
• You are charged only for live rooms actually saved. Offline creators, empty searches, duplicates, filtered-out rooms, and the run summary are all free.

Why choose this Actor

Most TikTok LIVE scrapers do one thing: you give them a keyword, they give you rooms. This one answers four different questions in a single run:

Question	How
"What's live and trending right now?"	trending mode — no keyword needed
"Is this creator live?" (for a whole list of creators)	creator mode
"Who's live about this topic?"	keyword mode
"Which of these rooms actually deserves my attention?"	built-in priority scoring, in every mode

Quick start — 3 steps

1. Leave every discovery field empty.
2. Set Max live rooms per source to 50 for a fast, cheap first run.
3. Click Start. You get ~50 scored trending live rooms in about a minute.

No input is required at all. When you want something more specific, fill in creators or keywords — there is no mode to switch.

How discovery works

What you fill decides what runs. There is no mode selector to keep in sync with your inputs.

What you fill	What runs	Best for
Nothing	Trending — hot/active rooms across gaming, music, chat, dance, cooking, sports, beauty, comedy and more	Broad monitoring, "what's happening now"
Creators to check	A liveness check per creator. Full room data when live, a clear NOT_LIVE row when offline	Creator watchlists, go-live tracking, agency monitoring
Keywords to search	One search per keyword	Niche tracking, markets, games, interest areas
Both	Both, in one run	Full LIVE intelligence sweeps
Either, plus "Also discover trending rooms"	Adds a trending sweep on top	Broad coverage alongside a watchlist

Two things follow from this that are worth knowing:

• A trending-only run needs no input. Press Start. The "Also discover trending rooms" toggle is only for adding trending alongside creators or keywords.
• No combination produces an empty run. With nothing filled you get trending, so there is no way to configure a run that has nothing to do.

Input reference

Parameter	Type	Default	Description
creatorProfiles	string[]	[]	Creators to check — @name, name, or profile URL. Filling this runs creator discovery
keywords	string[]	[]	Keywords to search. Filling this runs keyword discovery
includeTrending	boolean	false	Add a trending sweep alongside creators/keywords. Not needed for a trending-only run
trendingTerms	string[]	[]	Override the built-in trending seed terms
minViewerCount	integer	0	Minimum current viewers. Filtered rooms are not charged
sortBy	string	priorityScore	priorityScore, viewerCount, or default
maxRooms	integer	300	Cap per discovery source
maxRoomsPerKeyword	integer	50	Cap applied to each keyword separately
maxTotalRooms	integer	1000	Absolute ceiling for the whole run — the hard cost cap
includeCreatorFields	boolean	true	Include creator-level fields
includeActivitySignals	boolean	true	Include audience/engagement fields
includeStreamUrls	boolean	true	Include playback URLs, quality, resolution and expiry
includeSummary	boolean	true	Save the run summary to the key-value store
deduplicateRooms	boolean	true	Drop rooms found by more than one route, before billing
proxyConfiguration	object	Apify Proxy	Proxy settings

Cost controls — how the three caps interact

They are applied in this order, and the total always wins:

maxRoomsPerKeyword   ->  applied to each keyword on its own
maxRooms             ->  applied to each discovery source (trending counts as one)
maxTotalRooms        ->  absolute ceiling for the entire run

Worked example. Mode ["trending","keyword"], 5 keywords, maxRooms: 300, maxRoomsPerKeyword: 50, maxTotalRooms: 1000:

• trending can contribute up to 300 rooms
• each keyword can contribute up to 50 → up to 250
• so the run saves at most 550 rooms, and would stop at 1,000 regardless

Without maxTotalRooms, the same run with maxRoomsPerKeyword raised to 300 could save 1,500 rooms. That is why the total cap exists and defaults to a finite value.

Sample input — trending discovery

{
  "maxRooms": 50,
  "maxTotalRooms": 100,
  "sortBy": "priorityScore",
  "proxyConfiguration": { "useApifyProxy": true }
}

Sample input — creator monitoring + keyword search

{
  "creatorProfiles": [
    "https://www.tiktok.com/@aljazeera",
    "@giulianasanes",
    "elisefvc"
  ],
  "keywords": ["gaming", "asmr"],
  "maxRoomsPerKeyword": 50,
  "maxTotalRooms": 500,
  "minViewerCount": 100,
  "sortBy": "priorityScore",
  "proxyConfiguration": { "useApifyProxy": true }
}

Sample input — gaming vertical via custom trending terms

{
  "trendingTerms": ["valorant", "roblox", "minecraft", "fortnite"],
  "maxRooms": 200,
  "minViewerCount": 50,
  "proxyConfiguration": { "useApifyProxy": true }
}

Output

Each record is one public TikTok LIVE room with a flat, CSV-friendly schema (51 fields), available in five dataset views:

• Overview — key metrics and monitoring buckets at a glance
• Live Rooms — full room details with engagement and activity signals
• Creators — creator-focused view with follower and profile data
• Playback — stream URLs with quality, resolution and expiry
• Monitoring — qualification view with priority scores, buckets, and reasons

The Overview tab in the Apify console. Switch tabs with the view selector above the table.

Full example record

{
  "sourceMode": "trending",
  "sourceInputType": "live_discovery",
  "sourceKeywords": [],
  "sourceCreators": [],
  "scrapedAt": "2026-08-18T02:05:20.870Z",
  "roomId": "7675172478674668305",
  "roomUrl": "https://www.tiktok.com/@pagel.lover6/live",
  "liveTitle": "We have alot of fun for you",
  "roomDescription": "Pagel In Love",
  "isActive": true,
  "startTime": "2026-08-18T01:09:34.000Z",
  "minutesLive": 55,
  "coverImageUrl": "https://p16-common-sign.tiktokcdn.com/...",
  "snapshotImageUrl": "https://p19-common-sign.tiktokcdn.com/...",
  "categoryLabel": null,
  "categoryId": null,
  "gameTags": [],
  "creatorUsername": "pagel.lover6",
  "creatorProfileUrl": "https://www.tiktok.com/@pagel.lover6",
  "creatorId": "7666074142324688000",
  "creatorDisplayName": "Pagel Lover",
  "creatorAvatarUrl": "https://p16-common-sign.tiktokcdn.com/...",
  "creatorVerified": false,
  "creatorFollowersCount": 692,
  "creatorFollowingCount": 30,
  "creatorBio": "Pagel In Love",
  "viewerCount": 790,
  "totalUserCount": 13543,
  "viewersPerMinute": 246,
  "likeCount": 16120,
  "commentCount": 176,
  "shareCount": 16,
  "hasCommerce": false,
  "streamUrlHls": "https://pull-hls-f16-sg01.tiktokcdn.com/...",
  "streamUrlFlv": "https://pull-f5-sg01.tiktokcdn.com/...",
  "streamUrlRtmp": "https://pull-f5-sg01.tiktokcdn.com/...",
  "streamQuality": "hd",
  "streamResolution": "960x1920",
  "streamUrlExpiresAt": "2026-09-01T01:58:46.000Z",
  "streamVariants": [
    { "quality": "hd", "flvUrl": "https://...", "hlsUrl": "https://..." },
    { "quality": "ao", "flvUrl": "https://...", "hlsUrl": null }
  ],
  "priorityScore": 65,
  "priorityReason": "Active room, Moderate viewer count, High engagement, Trending discovery",
  "audienceBucket": "medium",
  "activityBucket": "high",
  "isTrendingCandidate": true,
  "monitoringBucket": "watch-now",
  "monitoringReason": "Currently active with meaningful engagement",
  "position": 1,
  "isPartial": false,
  "errorType": null,
  "errorMessage": null
}

One sample record per dataset view

Every view is a projection of the same row, so a field means the same thing in all of them. URLs are shortened below for readability; real output carries the full signed URL.

Overview tab

Key metrics and monitoring buckets at a glance — the view to scan first.

{
  "sourceMode": "trending",
  "roomId": "7675172478674668305",
  "roomUrl": "https://www.tiktok.com/@pagel.lover6/live",
  "liveTitle": "We have alot of fun for you",
  "minutesLive": 55,
  "isActive": true,
  "creatorUsername": "pagel.lover6",
  "viewerCount": 790,
  "viewersPerMinute": 246,
  "likeCount": 16120,
  "commentCount": 176,
  "hasCommerce": false,
  "priorityScore": 65,
  "audienceBucket": "medium",
  "monitoringBucket": "watch-now",
  "scrapedAt": "2026-08-18T02:05:20.870Z"
}

Live Rooms tab

Full room detail with engagement, category and commerce signals.

{
  "roomId": "7675172478674668305",
  "roomUrl": "https://www.tiktok.com/@pagel.lover6/live",
  "liveTitle": "We have alot of fun for you",
  "roomDescription": "Pagel In Love 🤤",
  "isActive": true,
  "startTime": "2026-08-18T01:09:34.000Z",
  "minutesLive": 55,
  "coverImageUrl": "https://p16-common-sign.tiktokcdn.com/...",
  "snapshotImageUrl": "https://p19-common-sign.tiktokcdn.com/...",
  "categoryLabel": null,
  "categoryId": null,
  "gameTags": [],
  "creatorUsername": "pagel.lover6",
  "viewerCount": 790,
  "totalUserCount": 13543,
  "viewersPerMinute": 246,
  "likeCount": 16120,
  "commentCount": 176,
  "shareCount": 16,
  "hasCommerce": false,
  "priorityScore": 65,
  "priorityReason": "Active room, Moderate viewer count, High engagement, Trending discovery",
  "sourceMode": "trending",
  "position": 1
}

Creators tab

Creator-focused view for watchlists and outreach shortlists.

{
  "creatorUsername": "pagel.lover6",
  "creatorProfileUrl": "https://www.tiktok.com/@pagel.lover6",
  "creatorId": "7666074142324688000",
  "creatorDisplayName": "Pagel Lover 🤤",
  "creatorAvatarUrl": "https://p16-common-sign.tiktokcdn.com/...",
  "creatorVerified": false,
  "creatorFollowersCount": 692,
  "creatorFollowingCount": 30,
  "creatorBio": "Pagel In Love 🤤",
  "isActive": true,
  "liveTitle": "We have alot of fun for you",
  "viewerCount": 790,
  "priorityScore": 65,
  "sourceMode": "trending"
}

Playback tab

Stream URLs with quality, resolution and expiry. Check streamUrlExpiresAt before using a URL.

{
  "roomId": "7675172478674668305",
  "creatorUsername": "pagel.lover6",
  "liveTitle": "We have alot of fun for you",
  "isActive": true,
  "viewerCount": 790,
  "streamQuality": "hd",
  "streamResolution": "960x1920",
  "streamUrlHls": "https://pull-hls-f16-sg01.tiktokcdn.com/...",
  "streamUrlFlv": "https://pull-f5-sg01.tiktokcdn.com/...",
  "streamUrlRtmp": "https://pull-f5-sg01.tiktokcdn.com/...",
  "streamUrlExpiresAt": "2026-09-01T01:58:46.000Z"
}

Monitoring tab

Qualification view — scores, buckets, reasons, and any extraction errors.

{
  "roomId": "7675172478674668305",
  "liveTitle": "We have alot of fun for you",
  "creatorUsername": "pagel.lover6",
  "isActive": true,
  "viewerCount": 790,
  "viewersPerMinute": 246,
  "minutesLive": 55,
  "likeCount": 16120,
  "priorityScore": 65,
  "priorityReason": "Active room, Moderate viewer count, High engagement, Trending discovery",
  "audienceBucket": "medium",
  "activityBucket": "high",
  "isTrendingCandidate": true,
  "monitoringBucket": "watch-now",
  "monitoringReason": "Currently active with meaningful engagement",
  "isPartial": false,
  "errorType": null,
  "errorMessage": null
}

Creator offline — example NOT_LIVE record

When a creator you asked about is not currently live, you get an explicit row rather than silence. This row is not charged.

{
  "sourceMode": "creator",
  "sourceInputType": "creator_profile",
  "sourceCreators": ["examplecreator"],
  "scrapedAt": "2026-08-18T09:14:02.551Z",
  "roomId": null,
  "roomUrl": null,
  "liveTitle": null,
  "isActive": false,
  "creatorUsername": "examplecreator",
  "viewerCount": null,
  "streamUrlHls": null,
  "priorityScore": 0,
  "priorityReason": "Error during extraction",
  "monitoringBucket": "low-priority",
  "monitoringReason": "@examplecreator is not currently live",
  "isPartial": true,
  "errorType": "NOT_LIVE",
  "errorMessage": "@examplecreator is not currently live"
}

Error rows carry the same columns as a live-room row, so a CSV export keeps a stable header.

Output fields

Source lineage fields (5)

Field	Description
sourceMode	Discovery mode that found this room (trending, creator, keyword)
sourceInputType	Input type (live_discovery, creator_profile, keyword_search)
sourceKeywords	Keywords used in this discovery (array)
sourceCreators	Creator usernames targeted (array)
scrapedAt	ISO 8601 timestamp of extraction

Room identity fields (12)

Field	Description
roomId	TikTok LIVE room ID
roomUrl	Direct URL to the live room
liveTitle	Stream title set by the creator
roomDescription	Room description (usually the creator's bio)
isActive	Whether the room is currently active
startTime	Stream start time (ISO 8601)
minutesLive	How long the room had been live when it was scraped
coverImageUrl	Static cover/thumbnail image
snapshotImageUrl	A frame captured from the stream as it is now
categoryLabel	Room category or tag (e.g. Gaming)
categoryId	Numeric category/game tag ID
gameTags	Game tags on the room (array)

Creator fields (8)

Field	Description
creatorUsername	TikTok username
creatorProfileUrl	Full profile URL
creatorId	TikTok user ID
creatorDisplayName	Display name / nickname
creatorAvatarUrl	Profile picture URL
creatorVerified	Verified badge status
creatorFollowersCount	Follower count
creatorFollowingCount	Following count
creatorBio	Profile bio

Audience / activity fields (7)

Field	Description
viewerCount	Viewers watching right now
totalUserCount	Cumulative viewers for the whole session
viewersPerMinute	Audience accumulation rate — see Reading momentum below
likeCount	Like count on the live session
commentCount	Comment count
shareCount	Share count
hasCommerce	Whether the room has shopping/commerce enabled

Playback fields (7)

Included when includeStreamUrls is on.

Field	Description
streamUrlHls	HLS playback URL for the best available quality
streamUrlFlv	FLV playback URL for the best available quality
streamUrlRtmp	RTMP pull URL
streamQuality	Quality key of the selected stream (e.g. hd)
streamResolution	Stream dimensions, e.g. 960x1920
streamUrlExpiresAt	When these URLs stop working (ISO 8601)
streamVariants	Every available quality with its FLV/HLS URLs (nested array)

Discovery / qualification fields (7)

Field	Description
priorityScore	Composite priority score (0–100)
priorityReason	Human-readable explanation for the score
audienceBucket	small / medium / large based on viewer count
activityBucket	low / medium / high based on engagement signals
isTrendingCandidate	Whether the room qualifies as trending
monitoringBucket	watch-now / watchlist / low-priority
monitoringReason	Explanation for the monitoring classification

Run metadata fields (4)

Field	Description
position	Rank of the row in the delivered, sorted results
isPartial	Whether the record has incomplete data
errorType	Why a row carries no room: NOT_LIVE, NO_ROOMS, or REQUEST_FAILED
errorMessage	Error message if extraction failed

Reading momentum

Two derived fields separate a room that is big from a room that is taking off. No other TikTok LIVE scraper in the Apify store returns anything derived.

Field	Meaning
minutesLive	Minutes between the stream starting and this row being scraped
viewersPerMinute	totalUserCount ÷ minutesLive — how fast the room is pulling viewers

Why it matters: two rooms can both show 5,000 cumulative viewers, but one reached that in 10 minutes and the other took 10 hours. The first is worth opening now.

Room A   totalUserCount 5,000   minutesLive  10   ->  viewersPerMinute 500   <- surging
Room B   totalUserCount 5,000   minutesLive 600   ->  viewersPerMinute   8   <- background noise

Note the denominator: the rate is built on totalUserCount (everyone who has passed through the room), not on viewerCount (who is watching right this second). Dividing a concurrent count by elapsed time would not mean anything.

Both fields are null when the room has been live for under a minute — the denominator is too small to say anything useful, and a burst of early joins would otherwise report an implausible rate. They are also null when TikTok does not expose a start time.

Sort by viewersPerMinute to build a "what's catching fire right now" feed; sort by viewerCount for "what's biggest right now". They are different questions.

Working with playback URLs

The playback URLs are signed and time-limited. Always check streamUrlExpiresAt before using one — after that timestamp the URL returns an error, and the fix is to re-run the Actor rather than to retry the old URL.

• streamUrlHls is the most broadly compatible (players, browsers, ffmpeg).
• streamUrlFlv is what TikTok's own web player uses.
• streamVariants lists every quality, including ao (audio-only). The flat streamUrl* fields always point at the best video quality, never at audio-only.

This Actor does not record streams — but it hands you everything a recorder needs.

Recording handoff — worked example

streamUrlHls opens directly in ffmpeg, VLC, or any HLS player. To archive a single room:

# -c copy remuxes without re-encoding, so this is cheap and lossless.
# It runs until the creator ends the stream, or until you stop it.
ffmpeg -i "<streamUrlHls>" -c copy "room-<roomId>.mp4"

To record every live room a run found, straight from the dataset:

DATASET_ID="<your run's dataset id>"

curl -s "https://api.apify.com/v2/datasets/$DATASET_ID/items?view=playback&format=json" \
  | jq -r '.[] | select(.streamUrlHls != null and .isActive == true)
           | [.roomId, .streamUrlHls] | @tsv' \
  | while IFS=$'\t' read -r roomId url; do
      echo "recording $roomId"
      ffmpeg -nostdin -loglevel warning -i "$url" -c copy "room-$roomId.mp4" &
    done
wait

Three things to get right:

1. Start promptly. The URLs are signed and expire — check streamUrlExpiresAt. If a recording fails to start, re-run the Actor for fresh URLs rather than retrying the old ones.
2. Filter to isActive: true. A room that ended between discovery and playback has a URL that resolves to nothing.
3. Cap your concurrency. The snippet above backgrounds every room at once; for a 300-room run, batch it instead.

For a managed alternative, feed streamUrlHls into a dedicated recorder Actor — this Actor is the discovery half of that pipeline, not a replacement for it.

Run summary

When includeSummary is enabled, an aggregated summary is written to the key-value store under the key SUMMARY — not to the dataset. It is not charged.

It lives outside the dataset deliberately: a summary carries aggregates, arrays and nested objects, so appending it to a dataset of flat room rows breaks the "one row = one live room" contract and corrupts a straight CSV export.

Find it under the run's Storage → Key-value store → SUMMARY, or via the API at /key-value-stores/{id}/records/SUMMARY.

{
  "recordType": "summary",
  "totalRoomsReturned": 68,
  "totalRoomsAfterDedup": 68,
  "roomsBySourceMode": { "trending": 68 },
  "roomsWithStreamUrl": 68,
  "roomsWithCommerce": 3,
  "activeRoomsByAudienceBucket": { "medium": 12, "small": 45, "large": 11 },
  "roomsByMonitoringBucket": { "watch-now": 67, "low-priority": 1 },
  "topCreators": [{ "username": "elisefvc", "roomCount": 1 }],
  "averageViewerCount": 618,
  "topLiveTitles": ["Tribunnews Live 24 Jam", "Watch GB News live"],
  "runStatus": "completed",
  "startedAt": "2026-03-24T12:06:28.461Z",
  "finishedAt": "2026-03-24T12:07:49.199Z",
  "durationSeconds": 81
}

Field availability

Not every field is available for every room. This table reflects what the Actor returns today.

Availability	Fields
Always present	sourceMode, sourceInputType, scrapedAt, roomId, roomUrl, isActive, startTime, coverImageUrl, snapshotImageUrl, creatorUsername, creatorProfileUrl, creatorId, creatorDisplayName, creatorAvatarUrl, creatorFollowersCount, creatorFollowingCount, totalUserCount, likeCount, commentCount, shareCount, hasCommerce, streamUrlFlv, streamUrlRtmp, streamQuality, streamResolution, streamUrlExpiresAt, streamVariants, priorityScore, priorityReason, audienceBucket, monitoringBucket, monitoringReason, position
Usually present	liveTitle, viewerCount, roomDescription, creatorBio, streamUrlHls
Only when applicable	categoryLabel, categoryId, gameTags — rooms carrying a hashtag or game tag
Rarely populated	creatorVerified — the live-search surface does not expose the badge, so it reads false

Unavailable fields are null rather than omitted, so the column set stays stable across runs and exports.

Priority scoring

Each room receives a composite score (0–100):

Signal	Points
Active room	+20
Very high viewers (10,000+)	+30
Solid viewers (1,000+)	+25
Moderate viewers (100+)	+15
High engagement (10,000+ total)	+20
Large creator (1M+ followers)	+15
Mid-size creator (100K+)	+10
Trending discovery	+10
Keyword/creator discovery	+5

priorityReason names the factors that contributed, so the score is auditable rather than magic.

Monitoring buckets

Bucket	When assigned
watch-now	Active room with score 60+, medium/large audience, or high activity
watchlist	Score 40+ or 50+ viewers
low-priority	Low signals or insufficient data

Use cases

• LIVE trend discovery — what is trending on TikTok LIVE right now
• Creator monitoring — track when creators go live, at list scale
• Audience research — public engagement signals from live rooms
• Niche monitoring — LIVE activity for specific topics or markets
• Live-commerce research — find rooms with shopping enabled via hasCommerce
• Recording pipelines — hand streamUrlHls to a downstream recorder
• AI/analytics workflows — structured LIVE data for dashboards, alerts, and LLM analysis

Pricing & billing

This Actor is pay-per-event, with two charges depending on how you discover rooms:

Mode	What you pay for
trending / keyword	Each live room saved to the dataset
creator	Each creator checked, live or not — the room data is then included, not charged again

Creator mode is billed per check because the work is per creator: the Actor loads a profile whether or not that creator turns out to be live. Billing that way means your cost is predictable from your input list before you press Start. The current rates are shown on the Actor's Pricing tab and in the Apify Console — this README deliberately does not quote figures, because rates change and the Console is always authoritative.

What you are charged for

	Charged?
A live room found by trending or keyword	Yes — one result
A creator checked in creator mode	Yes — one check, live or not
A live room found via creator mode	Free — covered by the check
The same creator retried after a failure	Free — checks are deduplicated per run
A search that returns nothing (NO_ROOMS row)	Free
A request that failed all retries (REQUEST_FAILED row)	Free
An offline creator's NOT_LIVE row itself	Free — the check is the charge, not the row
A duplicate room found by a second discovery route	Free — dropped before billing
A room excluded by minViewerCount	Free — filtered before billing
Rooms beyond maxTotalRooms	Free — never collected
The run summary	Free — and stored outside the dataset
Playback URLs, commerce flags, creator enrichment	Free — included in the row price

In short: trending and keyword bill for rooms you receive; creator mode bills for creators you asked us to check.

Estimating a run

Because billing is per saved room, the row count is the bill. Use the caps to bound it:

Run	Rooms saved (max)
Trending, maxRooms: 50	50
Trending, maxRooms: 300	300
5 keywords, maxRoomsPerKeyword: 50	250
Mixed (trending 300 + 5 keywords × 50)	550
Any run with maxTotalRooms: 1000	1,000, whatever else is set
200 creators in creator mode	200 checks — the count of creators you submitted

Set maxTotalRooms to the largest number of rooms you are willing to pay for in one run, and the run cannot exceed it.

For creator-only runs, also see Running creator checks cheaply below — selecting 1 GB of memory cuts the platform cost of those runs by 4×.

Running creator checks cheaply — set memory to 1 GB

Creator checks are browserless: the Actor reads TikTok's server-rendered page data over plain HTTP, so no browser is started. Trending and keyword discovery still need one, which is why the default memory is 4 GB.

If your run only checks creators, set memory to 1 GB in the run options. Apify bills memory × runtime, so this is a straight 4× saving on those runs, and the checks are unaffected — a creator-only run never launches a browser at all.

Run contains	Recommended memory
Creators only	1 GB
Any trending or keyword discovery	4 GB (the default)

Running trending or keyword at 1 GB is not recommended: the browser is starved of CPU and can be killed mid-navigation.

Spending limits

The Actor respects the per-run spending limit set in the Apify Console. When the limit is reached it stops collecting gracefully — results already saved stay in the dataset and the run finishes normally rather than erroring.

Authentication, proxy & reliability

No TikTok login, cookies, or API key are required. You never supply credentials, and the Actor reads only public LIVE surfaces.

Proxy use is recommended for real runs — TikTok rate-limits unproxied traffic.

🚦 Proxy policy

Use Apify Datacenter proxy (the default) or no proxy. Both work reliably for TikTok LIVE discovery at this actor's conservative concurrency, and Datacenter is what you get by simply leaving the proxy field alone:

{
  "proxyConfiguration": { "useApifyProxy": true }
}

Apify Residential proxy is not supported. The run stops immediately if apifyProxyGroups includes RESIDENTIAL. The reason is structural rather than technical: this Actor is pay-per-event, so Apify platform usage — including proxy bandwidth — is billed to the developer, not to you. This Actor's page loads are measured in megabytes, and residential bandwidth costs an order of magnitude more than datacenter, so a single residential run can cost several times its own revenue. Blocking it is what keeps the per-result price low for everyone else.

If you genuinely need residential routing, supply your own provider through the proxy editor's Custom proxy URLs field. That traffic goes through your provider and your account, so it is unaffected by this policy:

http://user:pass@proxy.iproyal.com:12321
http://user:pass@proxy.brightdata.com:22225
http://user:pass@proxy.oxylabs.io:7777

Reliability behaviour worth knowing:

• Failed pages are retried with exponential backoff and jitter, up to 4 attempts.
• Auth and billing failures (401, 402, 403) are not retried — they never succeed.
• If a page still fails, the run logs result set is TRUNCATED with the counts, so a short result is never mistaken for "TikTok had no more rooms".
• A failing trending term is isolated — the other terms continue.
• Partial records are marked isPartial: true rather than dropped.

Limitations

• All data comes from public TikTok LIVE surfaces. No private data, login-required features, or private analytics.
• Playback URLs expire. They are signed and time-limited — check streamUrlExpiresAt. Re-run the Actor to get fresh URLs.
• No country/region filter. TikTok's public live-search surface does not expose a region for rooms, so no reliable geo filter is possible and none is offered.
• Live rooms are time-sensitive. Rooms may end between discovery and extraction.
• maxRooms is a cap, not a guarantee. TikTok may return fewer active rooms than requested, and results vary through the day.
• Some fields are source-dependent. See the field availability table above.
• Viewer counts may be approximate. TikTok may expose rounded or delayed counts.
• Prioritization fields are heuristics. Scores and buckets are derived from public signals — practical guidance, not exact rankings.
• Keyword mode depends on TikTok's search surface supporting live-room results, which varies by region and availability.

What this Actor is NOT

• Not a LIVE video recorder — it returns playback URLs, not video files. Pair it with a recorder if you need the footage.
• Not a real-time chat or gift listener — it takes a snapshot per run, it does not stay attached to a stream.
• Not a continuous monitor — schedule repeat runs instead (see below).
• Not a private-data scraper — public surfaces only.

Common mistakes

Mistake	What happens	Fix
Expecting a trending sweep alongside keywords	Keywords alone run keyword discovery only	Turn on Also discover trending rooms
Turning on Also discover trending rooms for a trending-only run	Harmless, but unnecessary — trending already runs when nothing else is filled	Leave it off
Confusing Keywords to search with Trending seed terms	Keywords each run their own capped search; seed terms steer one trending sweep	Use Keywords for topics you want searched; seed terms only to redirect trending
Expecting maxRooms to cap the whole run	Each source gets its own budget	Set maxTotalRooms
Setting minViewerCount high on a quiet niche	Few or no rows returned (and nothing charged)	Lower the threshold, or set it to 0
Storing playback URLs for later	They expire — see streamUrlExpiresAt	Use them promptly, or re-run for fresh URLs
Looking for the summary in the dataset	It is in the key-value store under SUMMARY	Storage → Key-value store

Troubleshooting

Symptom	Likely cause	Fix
NO_ROOMS on every source	TikTok soft-blocked the session, or there genuinely are no matching live rooms	Enable Apify Proxy (Datacenter) and retry
A row with errorType: REQUEST_FAILED	That request timed out or errored on every retry, so it never ran. The result is missing, not empty	Re-run. If it persists, reduce the number of keywords per run
Log says result set is TRUNCATED	A page kept failing after all retries	Re-run; if it persists, lower concurrency or switch proxy group
Far fewer rooms than maxRooms	The cap is a ceiling, not a target	Normal. Try broader keywords or trending mode
NOT_LIVE for a creator you know is live	Creator status is a point-in-time snapshot	Re-run
Playback URL returns an error	The URL expired	Check streamUrlExpiresAt; re-run for fresh URLs
minViewerCount above 0 returns almost nothing	Rooms whose viewer count TikTok does not expose are excluded by the filter	Set minViewerCount: 0
Run stopped early with "maxTotalRooms reached"	Working as designed — the cost ceiling was hit	Raise maxTotalRooms
Run stopped with "spending limit reached"	The per-run spend limit in your Console was hit	Raise the limit in the Console
Run failed at startup mentioning residential proxy	RESIDENTIAL was selected in the proxy groups	Switch to Datacenter, or add your own provider under Custom proxy URLs — see Proxy policy
Run failed with "Nothing to run"	Only possible with a legacy saved input that still sends mode	Remove mode from the saved input and just fill the fields you want

Integrations & scheduling

Because LIVE is a moment-in-time surface, the usual pattern is a scheduled run:

1. Open the Actor → Schedules → create a schedule (for example every 15 minutes).
2. Point a webhook at your endpoint on ACTOR.RUN.SUCCEEDED to push new rooms into Slack, a sheet, or your own service.
3. Diff on roomId between runs to detect rooms that are newly live.

Results export as JSON, JSONL, CSV, Excel, HTML, or XML, and are reachable through:

• the Apify API (/datasets/{id}/items)
• the JavaScript and Python clients
• the Apify MCP server, for LLM and agent workflows
• Zapier, Make, and other Apify integrations

FAQ

Which discovery mode should I pick? There isn't one to pick. Fill creators, or keywords, or neither — the Actor runs the matching discovery. Leave everything empty for a trending sweep.

Do I need a TikTok account, login, cookies, or an API key? No. None of them. You only need an Apify account.

Am I charged if a creator is offline? Yes — creator mode bills per creator checked, not per creator found live, because the Actor loads the profile either way. The NOT_LIVE row itself is free; the check is the charge. This is what makes the cost predictable: it equals the number of creators you submitted.

Am I charged twice when a creator is live — once for the check and once for the room? No. Rooms found through creator mode are not charged as room results; the check covers them.

Am I charged if a search returns nothing? No. NO_ROOMS rows are free.

Am I charged twice if the same room appears in two modes? No. With deduplicateRooms on (the default), duplicates are removed before billing.

Am I charged for rooms filtered out by minViewerCount? No. Filtering happens before billing.

Can I play or download the stream? You get playback URLs (HLS/FLV/RTMP) that a player or ffmpeg can open. The Actor itself does not record.

Why did my playback URL stop working? It expired. Check streamUrlExpiresAt and re-run for fresh URLs.

Where is the run summary? In the key-value store under the key SUMMARY, not in the dataset — so your dataset stays one-row-per-room and exports cleanly to CSV.

Why did I get fewer rooms than I asked for? maxRooms is a cap, not a guarantee. TikTok's live inventory changes constantly.

Can I filter by country? No. TikTok's public live-search surface does not expose a region for rooms, so a reliable country filter is not possible and the Actor does not pretend to offer one.

How do I monitor creators continuously? Schedule the Actor to run on an interval and diff the results.

How do I keep a run cheap? Use one mode, set maxRooms to 50, and set maxTotalRooms to the maximum you are willing to pay for.

Changelog

Last verified: 2026-08-18

1.0

• Three discovery modes (trending, creator, keyword) with mixed-mode runs
• Playback stream URLs — HLS, FLV, RTMP with quality, resolution and an explicit expiry, plus all quality variants
• New fields — snapshotImageUrl, categoryId, gameTags, creatorAvatarUrl, creatorFollowingCount, totalUserCount, hasCommerce
• commentCount is now populated (it was always empty before)
• Removed permanently-empty fields — giftSignal, streamLanguage, creatorLikesCount, cohostCount. TikTok's live-search surface returns an empty link_mic stub on every room, so co-host state is not available from it
• Removed the countries input and region field — TikTok's live-search surface exposes no region, so the filter could never work; it is gone rather than misleading
• Run summary moved to the key-value store (SUMMARY), keeping the dataset one-row-per-room
• minViewerCount, sortBy and deduplicateRooms are now applied (previously accepted but not enforced)
• Cross-mode deduplication — a room found by two routes is saved and charged once
• New cost controls: maxRoomsPerKeyword and maxTotalRooms
• New trendingTerms input to steer trending discovery
• Apify Residential proxy is now rejected at startup — Datacenter, no proxy, and your own Custom proxy URLs are all supported. Keeps the per-result price sustainable.
• Removed the mode selector — discovery now follows the fields you fill, so the selector and the inputs can no longer disagree. Previously, filling a field whose mode was not selected dropped that input silently. Saved inputs that still send mode keep their old behaviour.
• New includeTrending toggle for adding a trending sweep alongside creators or keywords
• Removed the unused debugMode input
• A mode selected with an empty input now fails with a clear message instead of finishing as an empty successful run
• New creator-live-check charge — creator mode now bills per creator checked rather than per creator found live, so cost is predictable from your input list. Rooms found this way are no longer charged separately.
• Page-level retry with backoff, explicit truncation warnings, per-term failure isolation
• Filtered, duplicate and error rows are excluded from billing
• Five dataset views, including a new Playback view

Support

Found a bug, or need a field the Actor does not return yet? Open an issue on the Actor's Issues tab in the Apify Console with your run ID and input.