Meta Ad Library API free vs paid: what the free tier really costs

The free Meta Ad Library API charges nothing per call, but its spend, impression and demographic fields are reserved for political and social issue ads, it caps you at 200 calls per user token per hour, and it covers Meta only. For commercial competitor research, a paid feed is usually cheaper once you count engineering time.

That is the entire free versus paid decision compressed into a sentence, and the rest of this guide is the working behind it. The Ad Library API is a transparency instrument built for regulators, journalists and academic researchers, which is why its richest fields exist at all and why they stop where they do. When teams search for the Meta Ad Library API free versus paid, the question underneath is a procurement one: is the free tier enough to run competitor research, or is a paid feed the cheaper answer once your engineering time is priced in.

What the free Meta Ad Library API actually returns

The Ad Library API is the ads_archive endpoint on Meta's Graph API. It is the only official programmatic route into the Ad Library, and Meta charges no per call fee or subscription for it. Access, however, is not open. You create a Meta developer app, add the Ads Library API product, request the ads_read permission through App Review, and complete business and identity verification. Standard review runs five to ten business days; anything that reads as competitive intelligence rather than research can stretch to three or four weeks, per adlibrary.com's May 2026 review of the free tier.

The free tier in one definition: 200 calls per user access token per hour, creative text and delivery windows for almost any running Meta ad, spend and impression buckets reserved for political and issue ads, and no destination URL field at any access level. That definition is the whole decision in one line. Meta's own Graph API reference scopes ads_archive to political, electoral and social issue ads plus the US housing, employment and credit categories; third-party engineering write-ups verified against Graph API v20.0 reached the same conclusion in April 2026.

What comes back reliably, regardless of ad category, is a narrow set of identity and timing fields. Two of them function across nearly every ad type: page_id and ad_delivery_start_time. Alongside those you can pull page_name, publisher_platforms, the ad creative bodies, ad_creative_link_titles and ad_snapshot_url. Daily Intel Service's field-by-field breakdown (May 2026) notes that video and carousel ads frequently return several of the creative fields empty, because Meta only parses text it can extract cleanly.

Where the data is present but imprecise is spend and impressions. Both return as ranges, not exact numbers. A response might report a spend of lower_bound 1000 and upper_bound 5000, which means the advertiser spent somewhere between one and five thousand dollars. That is directionally useful for competitive monitoring and useless for financial modelling. The band on a large advertiser can be wide enough to swallow the entire signal.

Is the free Meta Ad Library API enough for competitor research?

For political, electoral and social issue advertising, yes. That is the job the API was built for, and it is the only place where you get the richer transparency fields like demographic and regional distribution. For an ordinary ecommerce, SaaS or app install campaign, the answer is usually no. The API withholds exactly the fields a competitor analysis needs, and it withholds them by design rather than by accident.

The gap is sharpest for commercial ads. The browser Ad Library shows a commercial ad to a human, but the API returns a much thinner record for the same page_id, and AdMapix's April 2026 engineering guide goes further: most commercial ads do not come back from ads_archive at all, because ad_type ALL means all within the political, issue and special category scope. Whatever side of that line your target sits on, the practical result is the same. You see the ad in the UI, you query it in code, and the fields you wanted are not there.

Europe is the one meaningful exception. Under the Digital Services Act, the European Union gets broader visibility into all ads than most other markets, which is why teams researching EU advertisers sometimes find the free API more generous than their US counterparts do. If your entire watchlist is Meta only, EU scoped, and you only need creative rotation rather than spend, the free tier can carry the workflow. For everything else, the architectural decision comes first, and it is covered in our guide to building on an ad intelligence API versus the Meta Ad Library API.

Where the free tier stops

Rate limits are the first wall. The published ceiling is 200 calls per user access token per hour, enforced as a rolling window, and it is per token rather than per app, so rotating tokens across a team buys you nothing. In practice you hit the limit earlier than the number suggests, because Meta throttles on an internal cost estimate rather than raw call count and signals it with error code 613, which many client libraries surface as a generic 400. A single search returning 10,000 results at 25 per page is 400 calls. Five competitor brands on an hourly schedule burns the budget before your own logic runs once.

The field ceiling is the second wall, and no query structure gets around it. There is no exact spend figure for a commercial ad. There are no creative media files, only a snapshot URL that points at a Meta-hosted preview page. There is no click through rate, no conversion data and no landing page or destination URL as a first class field. Meta has framed the Library as a disclosure instrument, so the omissions are policy rather than a gap waiting to be patched.

Platform coverage is the third wall. The API is Meta only: Facebook, Instagram, Messenger and the Audience Network. TikTok, YouTube, LinkedIn, Google, Snapchat and Pinterest are each a separate integration with its own access model, which is a real cost the moment your competitor set is not Meta exclusive.

Retention is the fourth, and it is the quiet one. Political and issue ads carry a multi-year retention commitment, commonly cited at around seven years. Non-political ads have a shorter window that Meta has never published, roughly weeks to months by working estimates, and ads that ran briefly, narrowly or were rejected for policy can disappear before an indexing pipeline catches them. A clean API result does not mean you have seen every ad a page ever ran; it means you have seen the subset that survived the window.

Token operations are the fifth. Long-lived user tokens expire after 60 days, so production pipelines need a scheduled refresh, and a server side integration usually means a System User token behind a verified Business Manager. None of this is hard, but all of it is work that recurs forever.

What free actually costs: the total cost of ownership

The free API is free of licence fees, not free of cost. The first bill arrives before you have made a single successful call. Pulling together the app review submission, the use case write up, the data handling explanation and the screen recording is commonly an eight to fourteen hour engineering task, and you spend it before you know whether your use case will be approved. A rejection resets the clock on a timeline you do not control.

The recurring bill is the maintenance surface. You own token refresh, pagination buffers sized under the rolling window, back off logic for error 613, deduplication across paginated pulls, normalisation of spend bands so you never mix midpoint and lower bound estimates in one dataset, and snapshot handling for any creative media you need. Every one of those is a line of code that can break in production at three in the morning and a person who has to fix it.

The tempting shortcut makes the comparison look better than it is. Teams that find the API too thin often fall back to scraping the Ad Library UI, which the Ad Library API terms prohibit for the official route and which breaks whenever Meta changes the page. After the Meta v. Bright Data ruling in the Northern District of California on 23 January 2024, scraping publicly accessible logged-out pages was held not to violate Meta's terms of service, and Meta dropped its appeal that February. That is persuasive district court authority, not a national precedent, and it does not authorise logged-in scraping or override GDPR obligations. It is also a maintenance commitment rather than a one time build, as we lay out in

our guide to automating Meta Ads Library scraping. The point for a cost decision is simple: a scraper you wrote is a scraper you maintain.

What a paid ad intelligence API adds

A paid feed sells three things the free tier cannot: breadth, depth and operations. Breadth is cross-platform coverage in one endpoint, so a single query returns Meta, TikTok, YouTube, LinkedIn and more instead of five integrations with five auth flows. Depth is the field set, which typically includes creative media rather than snapshot links, timeline history for pattern detection, enrichment that classifies angle and hook, and spend or engagement estimates for commercial ads. Operations is the part teams underprice: no app review, no token rotation, and a vendor who owns the breakage and carries a support commitment.

The scale argument is worth internalising because it explains the price. A continuously running index is not a weekend project. AdMapix's published creative index counted 60.9 million creative combinations active in a 30-day window from 9 August to 8 September 2026, with 43.2 million newly indexed in that same period. Buying access to coverage at that scale is fundamentally different from wiring a 200 call per hour endpoint to a spreadsheet, and the price of a subscription is really the price of not building and running that collection pipeline yourself.

That is also why managed providers exist in the middle. Pay as you go APIs and hosted scrapers remove the auth and maintenance burden without a full platform subscription, which is often the right first step for a pipeline that is not yet critical.

What paid ad intelligence APIs cost in 2026

Pricing has settled into a recognisable band. Meta's own first-party route to commercial ad data is the Content Library, accessed through the French secure data centre CASD, and it is researcher gated rather than commercial: as of January 2026 the SOMAR virtual data environment costs 371 dollars per team per month plus a one time 1,000 dollar project setup fee, with a four to eight week approval and restricted export. It is the right tool for academic work and the wrong tool for a brand reverse engineering a competitor.

Third-party pricing is more accessible and far more varied. Managed API providers sit at the low end per unit of data: ScrapeCreators checked on 1 September 2026 sells 25,000 credits for 47 dollars and 500,000 for 497 dollars, with most endpoints costing one credit and credits that do not expire. Specialist single-platform providers, such as SearchAPI for Reddit at 40 dollars per month for 10,000 searches, are worth it only when that platform is non-negotiable.

Managed browse-first tools bill a flat monthly subscription instead. Public 2026 pricing across the category clusters from about 49 dollars to 149 dollars per month, with Minea listed from 49, Foreplay at 59, PowerAdSpy at 69, AdSpy at 149, and adlibrary.com's own business tier at 329 euros per month for 1,000 plus credits across eight platforms with no app review. The range is wide because the products differ. Treat any single figure as the floor of a tier, not the price of the feature you actually need.

The comparison trap here is comparing only the lowest price per thousand requests. The number that matters is the returned field set, the platform coverage, whether pagination is honest, what the provider does with failed requests and how many separate systems your workflow has to keep alive. A cheap feed that costs you a day of engineering a month is more expensive than a subscription that just works.

Free vs paid: how do you decide?

Start with a checklist rather than a preference. The free Meta Ad Library API works when your research is Meta only, your watchlist is small, roughly ten pages and five countries or fewer, bucketed spend ranges are acceptable, app review delay does not block your timeline, you have a registered business entity for verification, and your sustained request rate stays around 100 to 150 calls per hour. It breaks when you need TikTok, YouTube, LinkedIn or Google in the same pipeline, when you monitor more than 20 to 30 advertisers on an hourly refresh, when you need exact spend or creative media files, or when your use case does not fit Meta's approved research categories.

The second test is arithmetic. Estimate the engineering hours the free path consumes per month, including the maintenance you will actually do rather than the maintenance you plan to do, multiply by a loaded hourly rate, and compare it against the subscription. For most teams running competitive monitoring across a real competitor set, the engineering line crosses the subscription price quickly, and that crossover is the answer to the free versus paid question. A useful pattern map makes the load concrete: a daily snapshot for five to ten pages is under 100 calls a day and fits the free tier comfortably; hourly monitoring of 200 plus advertisers needs 400 to 600 calls an hour and does not.

A third test is fit rather than cost. If your work is political transparency, compliance monitoring or academic research, the free API is not a compromise, it is the correct and often only sanctioned source, and provenance beats convenience. If your work is commercial competitor intelligence feeding a dashboard or an agent, the free tier is the wrong tool regardless of its price, and a paid or managed feed is the cheaper path once you count what you would build.

The hybrid most teams land on

In practice the clean answer is rarely all free or all paid. The common split is the official API for anything that must be Meta approved or Meta authoritative, and a managed or paid provider for the commercial and cross-platform coverage the free tier withholds. That keeps the sanction and provenance where it matters and buys breadth only where you need it. It also keeps a human in the loop, which is what turns a pile of scraped ads into a decision instead of noise.

If you are choosing a paid layer, our comparison of the

best ad intelligence tools for agencies is the place to match a provider to your workflow, and if the end consumer is an AI agent, connecting the Meta Ads Library API to AI agents covers the access model your agent will actually use.

Whichever side of the line you land on, the workflow that survives is the one that does not pretend to be complete. The free API is honest about its scope, and a good paid provider is honest about its coverage. The mistake is building a pipeline that looks comprehensive while quietly missing every ad that churned out of a retention window or every platform the endpoint never covered. Price the tool against that reality, and the free versus paid decision stops being about the invoice and starts being about what you can actually act on. For the monitoring workflow itself, see our

guide to monitoring competitor Facebook ads with the Meta Ads Library API.

Frequently asked questions