Threads Scraper — Posts, Profiles, Replies & Hashtags

Scrape Threads (threads.com) without a browser and without logging in.

Give it a keyword, a #hashtag, an @username, or any Threads URL. Get back flat rows ready for a spreadsheet, a dashboard, or an LLM pipeline.

---

Why this one

Threads scrapers break in two boring ways: they don't recognise the URL you gave them, and they die the moment Meta reshuffles the page. This one is built against both.

	This actor	Typical Threads scraper
How it finds data	Walks the page's own JSON for any thread_items node	Hard-coded path into a Relay wrapper Meta reshuffles at will
URL shapes understood	profile, /replies, /media, /reposts, /@user/post/CODE, /t/CODE, threads.net, ?si= tracking links	Often just the plain profile URL
Stable numeric author ID	✅ authorId on every row	Usually username only — which breaks when someone renames
One bad target	Skipped, with the reason; the rest still run	Whole run fails
What you pay for	Rows actually delivered, plus a $0.005 start fee	Start fee plus rows

Filtered, duplicate and failed rows are never billed. Each run carries a $0.005 start fee ($0.02 on the Apify free plan; charged per GB of memory, the default is 1 GB).

---

What you get

Real, unedited rows from this input:

{ "queries": ["@zuck", "#buildinpublic"], "profileTabs": ["posts", "replies", "media"], "maxItems": 25 }

→ 25 rows in one run: 1 profile + 24 unique posts, zero duplicates across the three tabs.

A post

{
  "recordType": "post",
  "code": "DVrwsE5EdSz",
  "url": "https://www.threads.com/@zuck/post/DVrwsE5EdSz",
  "text": "Meanwhile at Meta HQ",
  "authorUsername": "zuck",
  "authorId": "63055343223",
  "authorFullName": "Mark Zuckerberg",
  "authorIsVerified": true,
  "publishedAt": "2026-03-10T00:26:46.000Z",
  "likeCount": 7581,
  "replyCount": 1192,
  "repostCount": 232,
  "quoteCount": 56,
  "isReply": false,
  "mediaType": "image",
  "mediaCount": 1,
  "media": [{ "type": "image", "url": "https://instagram.fcdg4-1.fna.fbcdn.net/v/t51.82787-15/...", "width": 1080, "height": 1350 }],
  "hashtags": [],
  "sourceType": "profile",
  "sourceUrl": "https://www.threads.com/@zuck"
}

authorId is the account's stable numeric ID. Usernames get changed; this doesn't. Use it as your join key.

A profile row

{
  "recordType": "profile",
  "authorUsername": "zuck",
  "authorId": "63055343223",
  "authorFullName": "Mark Zuckerberg",
  "authorIsVerified": true,
  "followerCount": 5721349,
  "biography": "Mostly superintelligence and MMA takes",
  "bioLinks": [],
  "url": "https://www.threads.com/@zuck"
}

---

Input

Field	Type	Notes
queries	array	Required. Keywords, #hashtags, @usernames, or Threads URLs — mixed freely.
profileTabs	array	Which profile tabs to pull: posts, replies, media, reposts. More tabs = more posts.
expandThreads	boolean	Also open every post that has replies and scrape the conversation. Highest yield, highest cost.
maxItems	integer	Hard cap on delivered rows — your spend ceiling. 0 = no limit.
minLikes / minReplies	integer	Drop low-engagement posts. Filtered posts are not billed.
onlyVerified	boolean	Verified authors only.
excludeReplies	boolean	Top-level posts only.
mustHaveMedia	boolean	Only posts carrying an image or video.
postedAfter / postedBefore	string	YYYY-MM-DD date window.
includeProfileInfo	boolean	Emit the extra profile row shown above.
proxyConfiguration	object	Residential strongly recommended.

Every URL shape works

@zuck                                          https://www.threads.net/@zuck
https://www.threads.com/@zuck                  https://www.threads.com/@zuck/replies
https://www.threads.com/@zuck/media            https://www.threads.com/@zuck/post/CODE
https://www.threads.com/t/CODE                 https://www.threads.com/@zuck/post/CODE?si=xyz
#buildinpublic                                 ai agents

---

How much data you actually get — read this

Threads shows a limited number of posts to logged-out visitors, and this actor does not log in. Measured yield per page fetch:

Input	Posts per fetch
A post URL (post + its reply thread)	~31
Keyword search	~20
Hashtag	~18
Profile /replies tab	~8
Profile /media tab	~6
Profile main tab	~4

So a profile returns its most recent posts, not its full history. Scraping all three profile tabs is the way to widen that — measured ~18 unique posts for @zuck across three tabs versus 4 from the main tab alone — and expandThreads multiplies it further by pulling each post's replies.

If you need a complete profile archive going back months, this actor will not give you that, and no amount of retrying will change it. That is a Threads limit, not a bug, and it is stated here rather than discovered after you have paid.

---

Cost

You are billed $0.0025 per delivered row plus a $0.005 start fee per run ($0.02 on the Apify free plan; per GB of memory, the default 1 GB = one fee). Rows removed by your filters, duplicates, and failed pages are all free.

Job (one run, paid plan)	Rows	Cost
One profile, all three tabs	~18	~$0.05
One keyword search	~20	~$0.06
One post + full reply thread	~31	~$0.08
10 hashtags	~180	~$0.46

Set maxItems to cap any run. It is a hard ceiling.

---

Honest limitations

• No full profile history. See the table above. Deeper paging sits behind Threads' internal API and is deliberately not attempted.
• language is almost always empty. Threads does not populate detected_language for logged-out requests, so there is deliberately no language filter in the input — a filter that silently keeps everything is worse than no filter at all.
• Private and non-existent accounts return nothing. You get an explicit does not exist or has no public content message, and no rows are billed.
• Threads rate-limits. Each retry rotates to a fresh proxy session and re-establishes cookies, which is what actually clears a block — a fresh IP alone gets challenged again. If everything is blocked the run fails and bills no rows (only the start fee).
• A post with no replies yields one row. expandThreads only helps where conversation exists.

---

Why did my run fail?

A failed run always carries a plain-English reason in its status message (top of the run page, and the statusMessage field in the API) and never bills any rows — only the start fee applies. The four causes:

Status message starts with…	What it means	Fix
None of the … input value(s) could be understood	Every input was a URL shape the actor does not know. Threads URLs must be a profile, a profile tab (/replies, /media, /reposts), a post (/@user/post/CODE) or a short link (/t/CODE).	Check the URL. Anything that is not a Threads URL is treated as a keyword search, so bare words always work.
"postedAfter" must be a date like "2026-01-31"	A date filter was not an ISO date.	Use YYYY-MM-DD.
None of the requested targets exist … (HTTP 404)	Threads returned 404 for every target.	Check the spelling.
Threads rate-limited or refused every attempt	Threads blocked every fetch even after IP rotation — either an outright HTTP 429, or (under load) pages served empty right after one, which the actor retries and then reports as a block rather than as "no posts". Not caused by your input.	Wait a few minutes and retry. Avoid launching many runs at the same moment — bursts of parallel runs are what Threads rate-limits; put several targets in one run's queries instead, or space scheduled runs out. Raising maxRetries also helps.

A run with at least one good target succeeds and delivers what it could; the bad ones are listed in the run log and in the SUMMARY record of the key-value store.

---

Use it for

• Brand & competitor monitoring — watch a hashtag or keyword on a schedule.
• Creator research — recent posts, engagement counts, and verified status per account.
• Conversation mining — pull a viral post's full reply thread for sentiment or support signals.
• LLM pipelines — flat rows, no HTML, stable numeric IDs to join on.

---

Scheduling

Point the Apify Scheduler at this actor with a handful of keywords and postedAfter set to yesterday to build your own Threads history over time. Because billing is per delivered row plus a small start fee, and duplicates are removed within each run, a daily watch stays cheap.