How-To Guide
How much does the Threads API cost?
The Threads API has no paid tier or published price. See what Meta documents about access, rate limits and review, and what a tool's Threads cost covers.
By the Superdirector editorial team. Meta's official Threads developer documentation (overview, get-started, posts, keyword search, profiles, reference and changelog pages at developers.facebook.com) and every page cited here were read on November 14, 2026; vendor statements are as each page prints them on that date.
At a glance
How much does the Threads API cost?
Nothing in dollars, by documented absence. Meta's official Threads developer surface, read November 14, 2026 across its overview, get-started, posts, search and profile pages, publishes no paid tier, no per-call fee and no pricing page. Access is metered instead: a call budget of 4,800 times the account's impressions with a 48,000-call daily floor for a brand-new account, hard per-profile caps of 250 API-published posts, 1,000 replies, 100 deletions and 500 location searches per rolling 24 hours, and 2,200 keyword searches per user. The real costs are calendar costs: App Review for anyone outside your app's testers, a token refresh schedule that never ends, and media that must sit on a public server.
- Every cap on this page was read on Meta's own Threads documentation on November 14, 2026; Meta does not publish a price anywhere on that surface.
- The meter, not money: 4,800 times impressions per 24 hours (floor of 48,000 calls), 250 posts, 1,000 replies and 100 deletions per profile, 2,200 searches per user.
- The calendar bill: App Review with business verification for production access, 1-hour tokens that need a 60-day refresh job, and re-grants for private-profile users.
- Boundaries printed plainly: Superdirector does not publish to Threads; its plan terms cover Instagram, TikTok, YouTube and X.
The bill here is calendar time, not money. Step heights are editorial emphasis on how each stage stretches a launch, not measured durations; the two-round review budget is this article's planning advice, not a published Meta figure. Original diagram, hand-drawn from the get-started and App Review pages read November 14, 2026.
Make it work for your brand
Find it. Remake it. Post it. Superdirector turns the viral formats in your niche into your brand's carousels, X posts and video scripts, and keeps the calendar full.
What Meta publishes on price, and where the meter binds
Start with the absence itself, because it is the whole dollar answer. We checked Meta's Threads developer surface on November 14, 2026: the official overview, get-started guide, posts documentation, keyword search page, profiles page and the changelog carry no price, no tier and no billing model anywhere. The vendor pages that answer the same question agree on the reading: Blotato's guide states Meta publishes no paid tier, no per-call fee and no pricing page, and the Monocrawl comparison answers the cost question with Meta publishing no price and setting access by permissions and rate limits instead. What the official surface does publish is a meter, and that meter is the number to plan against.
| The Threads API meter | Official value (read on 2026-11-14) |
|---|---|
| General call budget | Calls within 24 hours = 4,800 x impressions; impressions floor 10, so 48,000 calls a day by arithmetic |
| API-published posts | 250 per profile per rolling 24 hours; a carousel counts as one |
| Replies | 1,000 per profile per rolling 24 hours |
| Deletions | 100 per profile per rolling 24 hours |
| Location searches | 500 per profile per rolling 24 hours |
| Keyword and topic-tag search | 2,200 queries per user per rolling 24 hours; empty results are free of the cap |
| Profile lookups | 1,000 requests per user per rolling 24 hours, public profiles with 100+ followers |
| CPU ceilings | 720,000 x impressions (total cputime) and 2,880,000 x impressions (total time) |
Two habits make the meter survivable. First, read your usage rather than guessing it: the threads_publishing_limit endpoint returns quota usage for posts, replies, deletions and location searches alongside its config objects, and reading those config fields at runtime is safer than hard-coding any of the numbers above. Second, remember the caps bind per profile, so 250 posts a day is generous for one account and still exactly 250 for a tool acting on behalf of many.
The calendar costs: App Review, tokens and the public-server rule
The money story is short, so the real planning question is how long access takes. While your app stays in development mode, accounts you invite as Threads testers can grant permissions immediately; to serve anyone without a role on your app, Meta's App Review page requires each permission to pass review first, and its reviewers test that the app actually uses those permissions before approving them. The submission itself has shape: Meta's review documentation lists a testable version of your app with instructions, a privacy policy and app details, and, for advanced access requested on or after June 12, 2023, business verification plus data handling questions. Blotato's page reports processing typically takes less than one week, often 2 to 3 days; hold that as a vendor-reported reading, because Meta's stored pages publish no turnaround figure. First-round rejections are a normal part of the route rather than a rarity, which is why the wise plan budgets two submissions, not one.
- Invite your own Threads account as a tester and build against it while the app stays in development mode; every endpoint works on tester accounts, and no review is needed for that.
- For production users, submit each needed permission: threads_basic for everything, threads_content_publish for posting, plus reply, insight, delete, search and discovery scopes as your job requires; expect the reviewer to test the app for real.
- Complete business verification and the data handling questions before requesting advanced access, since Meta's review flow asks for both.
- Start review early: one vendor-reported turnaround is under a week per round, and a rejection starts the round again, so two or three rounds of calendar time is the realistic budget.
Two standing obligations follow the review, and neither ever invoices. Tokens: Meta's get-started page states short-lived tokens are valid for 1 hour, long-lived tokens for 60 days, and permission grants from public-profile users are valid for 90 days with refresh extending them, while private-profile grants cannot be extended and the user must grant again. Media: the same page requires media used in publishing to sit on a publicly accessible server at the time of the attempt, because Meta's servers fetch it, so hosting is part of the wiring from day one.
The reading caps: search, lookups and what approval unlocks
If your job is reading rather than posting, the official pages carry the caps one by one. Keyword and topic-tag search allows 2,200 queries per user per rolling 24 hours, counted across all your apps for the same user, with queries that return no results free of the cap and sensitive or offensive keywords returning an empty array. Profile lookups cap at 1,000 requests per rolling 24 hours, and each lookup returns public profiles with at least 100 followers. Approval matters as much as the number: until the search permission clears App Review, a search covers only the authenticated user's own posts, and on standard access profile lookup reaches only some of Meta's official accounts, such as @meta and @threads, before the wider public surface opens.
Reading caps are also where secondhand numbers drift, so treat every table like this one: dated. One vendor guide still prints 500 keyword searches per rolling 7 days, a superseded figure, while Meta's own page states 2,200 per rolling 24 hours and the changelog records search limits changing more than once. The same date-discipline applies to vendor pricing: the Monocrawl page lists starting free allowances across the data-API market (its own from 150 credits, others in credits, records or units per their pages), but those are vendor-listed offers read on dated pages, not a card you can forecast against, and every one of them buys scraped public data rather than posting.
Publishing laws that trip builders, not budgets
A free API still finds ways to cost you failed runs. The most ordinary one is timing: Meta's posts page instructs waiting about 30 seconds before publishing a media container so the upload finishes processing, and a container that never publishes has by one vendor's reading an unpublished-container expiry after 24 hours. The rules that surprise people are written down: text posts cap at 500 characters with emojis counted by UTF-8 bytes, so an emoji-heavy post hits the ceiling sooner than a character counter suggests, and posts carry at most 5 unique links, a rule that has failed oversize posts with a named error code since December 22, 2025.
Capabilities have boundaries too, and they change the build more than any fee would. Meta's overview tells scheduling applications to enforce the publishing limits themselves, because no scheduling parameter exists; SocialAPI.ai's guide confirms there is no DM API and no messaging endpoints in the developer docs, and no way to edit a published post, so corrections mean delete and republish against the 100-a-day deletion cap. None of these is a dollar line. All of them are engineering lines, and the two-step container flow for images and video is where most of them surface: create the container, check its status, then publish.
The step that eats runs is the wait, not a fee. About 30 seconds is Meta's own instruction on its posts page; the 24-hour unpublished-container expiry is a vendor reading, so it is drawn dashed because Meta's stored pages do not print it. Original diagram, hand-drawn from the posts page read November 14, 2026.
Pay with patience, wire it yourself, or rent a pipe
Because Meta publishes no price, every dollar in this market is a vendor's dollar, and every honest comparison starts with what you would pay in effort. Measure your own job first: how many posts a day per profile, how many searches a day, whether anyone outside your tester circle must ever connect a Threads account. That answer picks the row in the table below, and the rows are ordered by how much infrastructure you own at the end of it.
| Your job | What the API gives you | What you still own |
|---|---|---|
| Prototype on your own account | Full endpoints as a tester, no fee, no review | OAuth wiring, token refresh job, public media hosting |
| Post for real users at 250 a day | No per-call fee, metered by the caps | App Review, business verification, re-permission flows, the publishing queue |
| Read public posts at scale | 2,200 searches and 1,000 lookups per day per user after approval | The approval round, then quota-aware polling; scrape vendors sit outside Meta's rules |
| Publish to Threads plus other networks | One more Graph-style surface with the same pattern | A second OAuth flow and another cap sheet per network, or a wrapper's subscription |
Three honest routes match those rows. Build direct when the API's own surface is your product: a monitoring or research tool lives on the reading caps and pays in review rounds rather than dollars. Rent a pipe when publishing is all you need: Blotato's page prices its plans from $29 to $499 a month with API access on paid plans, printed on its own page, and a wrapper absorbs the OAuth flow and the review wait for its own covered set. Hand over the calendar when the goal is a posting operation rather than a pipe: Superdirector (ours) runs that at $199 a month for one AI-run operated account, with posting permission standing at the account level, or $29 a month for MCP access plus hands-on creation, the version you run yourself. And the boundary this page will not blur: Superdirector's product publishes on Instagram, TikTok, YouTube and X, and Threads is not one of its covered platforms, so a Threads-direct decision stays with Meta's portal, this meter, and the review queue. On the model-in-the-loop layer above any API, one measured six-platform comparison found medians of $0.1020 versus $0.0981 per run with tokens varying far more between paths, a vendor's own measurement that prices the agent, not Meta's API.