best 20 social media api providers

pricing is graduated by account count: free for two, $6 each for three to ten, $3 up to a hundred, and $1 beyond that with no cap and no sales call. unlimited posts, analytics, messaging, comments, ads and webhooks are included at every tier. at fifty connected accounts that is an order of magnitude below the per-profile incumbents.

note the rebrand, because it makes most existing coverage stale: late became zernio in 2026 and getlate.dev now redirects. the sdks renamed too — the node and python packages and the cli all moved namespace — but api endpoints, keys and webhooks are unchanged, with a six-month grace period on the old packages.

x costs are passed straight through at x's own per-request rates rather than absorbed, which is now standard across aggregators and is a real budgeting trap if you post links.

the risk is company maturity. it is young, visibly founder-led, and has just been through a rebrand. that is a genuine concentration risk for something you build a product on, and it is the main reason to weigh buffer's stability against zernio's economics.

buffer reopened as a real developer platform: self-serve keys from your account settings, bearer auth, and endpoints for channels, posts, ideas and metrics. api access works on the free plan, which almost nothing else here offers.

the rate-limit design is the best in this ranking and it is worth copying. three rolling windows — fifteen minute, twenty-four hour and thirty day — and when you exceed one, the 429 response tells you which window you exhausted and includes a retry-after header. most first-party platform apis just say no.

the official mcp server documents support for claude, chatgpt, cursor, raycast and perplexity, and there is a cli. for agent work this is one of the more thoughtfully built offerings from an established vendor.

the ceiling is the catch. three thousand posts a month on free and fifteen thousand on team is low for a multi-tenant product, and analytics depth trails the specialists. higher limits exist by emailing them, which is not a plan you can budget against.

ayrshare covers more ground than anyone else: posts, analytics, comments, messages and ads across thirteen networks, with user profiles that abstract away per-platform oauth so you never register apps for each end user.

correct the price before you plan around it. software directories widely list this as starting around $25 a month; ayrshare's own pricing page says $149 for a single profile, $299 for up to ten and $599 for thirty, with per-profile overages above that. the cheap figure appears to be copied between listing sites.

rate limits are documented properly — three hundred requests per five minutes per user profile, with headers to track it — and unusually, they publish the consequence of ignoring them: a thousand 429s in twenty-four hours suspends the profile.

there are two mcp servers, one read-only for documentation and code generation and one for runtime actions. per-profile pricing is the structural problem: for a multi-tenant saas with many end users, this model gets expensive faster than almost any alternative here.

postiz is agpl-licensed and self-hostable, with a cloud tier from $29 a month. crucially the public api and webhooks are available on every tier including the cheapest — publer, metricool and sprout all gate api access behind their top plans.

there is an official mcp server, a node sdk and an n8n node, and development is very active.

two things to understand before committing. the agpl is a real constraint if you plan to build a commercial hosted product on a modified version — take advice rather than assuming. and self-hosting does not exempt you from platform-side approval: you still need your own tiktok audit, your own meta app review, your own linkedin partner status. that is the single most common misconception about self-hosted schedulers, and it is usually the hardest part of the whole project.

cloud channel caps are low relative to the graduated-pricing competitors, so at scale the self-hosted path is the one that makes sense.

the pricing is flat by tier rather than per account, which for a product connecting many profiles is a fundamentally different economic shape from ayrshare's.

there is a discrepancy worth knowing: blotato's pricing comparison table does not list api access as an included feature, which reads as though it is gated. the api documentation states plainly that api access is included on every paid plan, with only the free trial excluded. trust the docs — but be aware the pricing page will confuse your finance team.

the hosted mcp server is documented as working with claude code, claude desktop and cursor, and there are official n8n and make nodes that blotato itself recommends over raw api calls. rate limits are published at thirty requests a minute for both media upload and publishing.

the honest framing is that this is an ai content-generation product — ai writer, video, images — that has an api, rather than an api-first infrastructure company. credits are consumed by generation, which muddies cost prediction, and the documentation is thinner than ayrshare's.

create an app password and start calling. no app review, no business verification, no partner programme, no screencast. in a category where every other first-party platform gates access behind weeks of process, that is a genuine outlier.

writes use a points budget — creates cost three points, updates two, deletes one — against five thousand points an hour and thirty-five thousand a day, which works out to roughly 1,600 creates an hour. there are official typescript and go sdks from bluesky itself, plus community python and dart libraries.

the firehose is the thing nobody talks about enough. you can consume the entire network's activity in real time, for free. on every closed platform that capability either does not exist or costs enterprise money. for social listening agents this is unmatched.

the trade is reach. the audience is much smaller than x, instagram or tiktok, and because the protocol does not centralise engagement metrics the way closed platforms do, analytics are thin. every agent builder should start here while waiting on other platforms' approvals.

eleven platforms with publishing, comment and dm read-and-reply, review management, profile analytics and webhooks. the engagement coverage is more complete than most of the cheaper aggregators, which tend to stop at publishing.

pricing runs from free with ten posts a month through $17, $49 and $99 tiers up to a $699 enterprise plan, and api access, sdks and webhooks are included at every level. being able to build and test the real integration before paying is rare here.

mcp is supported alongside n8n and zapier.

it is early-stage, and that is the main caution. treat it as a higher-risk dependency than buffer or ayrshare, and check that the platform list still matches your needs before committing.

mixpost is self-hosted with a perpetual licence rather than a subscription. pro at $299 covers eleven networks with the rest api and an mcp server; enterprise at $1,199 adds white-label branding, customer dashboards and automated billing, which is genuinely a starter kit for reselling it as your own product.

unlimited social accounts and unlimited team members at every paid tier. against a per-profile subscription the arithmetic is not close over a couple of years.

the free lite tier only covers three networks, which makes it too limited to serve as a real evaluation — budget for the paid licence if you want to assess it properly.

as with postiz, self-hosting means you carry the operational burden and, more importantly, every platform approval. posting directly to tiktok requires passing tiktok's audit under your own client, which is the same wall you would face building from scratch.

pro at $100 a month covers ten thousand posts and five thousand comments with unlimited accounts and unlimited users; business at $400 raises the volumes tenfold. there are no per-seat or per-account fees at all.

for a multi-tenant product that is the decisive difference. per-profile pricing elsewhere scales linearly with your customer count and destroys margins; this does not.

api access is included even on the free tier, which covers twenty posts a month across three accounts — enough to build against.

x costs are handled as prepaid pass-through at x's own rates, stated openly rather than buried. fourteen-plus platforms including discord, slack, mastodon and google business, which is broader on the messaging side than most.

the current model is pay-per-usage: buy credits, they deduct per request, no contract and no minimum spend. that is a real improvement — you can start with a few dollars and there is no approval process. duplicate requests within a rolling twenty-four hour window are not charged twice, which is genuinely useful for polling design.

then there is the link tax. a post costs $0.015. a post containing a url costs $0.200. that is a thirteen-fold multiplier on exactly the thing most social automation does, and a thousand link posts a month is two hundred dollars in raw x fees before any aggregator subscription. every aggregator now passes this straight through. model it before you commit to anything.

reads are cheap individually — a tenth to a half a cent — but capped at two million post reads per billing cycle, above which you are into an enterprise conversation.

x now ships official mcp servers, and there is a detail worth catching: the action mcp covers search, user lookup, bookmarks, trends and article publishing, but it does not expose regular post creation. you can bookmark and publish articles through it; to actually tweet you still need the rest api. also worth knowing this is the third pricing regime since 2023, so budget for it changing again.

two hundred and fifty api-published posts and a thousand replies per rolling twenty-four hours, plus a hundred deletions and five hundred location searches. no other first-party publishing quota here comes close.

it covers single and carousel posts, media retrieval, profile management, reply moderation, mention handling, analytics, webhooks, keyword search and embeds. permissions are granular, so you request only what you use.

access friction is meaningfully lower than the rest of meta. threads was built with more developer-friendly onboarding and it shows.

the general call budget scales with impressions, which punishes low-reach accounts — a growth-stage account will bind on that long before it hits the posting quota. and the audience is smaller than instagram's, so it is rarely the only integration you need.

phyllo is not a posting api and should not be compared with the aggregators above on that basis. it aggregates consented creator data — the user authenticates, you read their authenticated metrics, including private ones no scraper can reach.

that consent model is the whole value. it is legally clean in a category where most alternatives are scraping vendors operating against platform terms, and for anything customer-facing or funded that difference matters more than price.

coverage spans youtube, tiktok, instagram, x, facebook, linkedin, snapchat and twitch among others, with social listening, screening, video intelligence and tiktok shop products alongside.

pricing is not published — the page offers customised plans and a quote form, with a free sandbox for evaluation. a $199 monthly figure circulates on software directories and does not appear on phyllo's own site. its closest competitor, insightiq, is a separate company in the same niche and worth evaluating alongside it.

register an application on any instance and go. oauth 2.0, no review process, and documentation that has been stable for years. in this category that combination is rare enough to be notable.

defaults are three hundred requests per five minutes per account and per ip, with media uploads and status deletions capped at thirty per thirty minutes, and standard rate-limit headers on every response.

the caveat that trips up most integrations: those are hardcoded defaults, and individual instance admins can configure different limits. your code has to read the headers and adapt rather than assuming the documented numbers.

federation is both the feature and the friction. there is no single api host, so supporting many instances means per-instance oauth apps, and an instance admin can rate-limit or defederate you. the audience is small.

there are two current apis: one where the user authenticates with instagram directly and no facebook page is involved, and one requiring the account linked to a page. both need an instagram business or creator account — personal accounts have no api access at all.

the basic display api was shut down at the end of 2024, which permanently killed personal-account access and a whole generation of feed-embed widgets. anything you read referencing it is dead information.

publishing is capped at a hundred api-published posts per rolling twenty-four hours per account, with carousels counting as one post against a separate fifty-carousel ceiling, and an endpoint to check your remaining budget.

the real cost is access. standard access only covers accounts you own; serving any third-party client needs advanced access, which means meta app review plus business verification plus a screencast demonstrating each permission in use. most applications are rejected first time, each round takes one to four weeks, and a realistic first build is four to eight weeks. that is why the aggregators above can charge what they charge.

correct the quota model first, because most published guidance is wrong about it. google's current documentation specifies a hundred search calls, a hundred upload calls, and ten thousand units per day for everything else. search and upload were carved out into their own daily allowances; the old arithmetic of dividing ten thousand units by the cost per search happens to give a similar answer for search and gets uploads badly wrong.

most list operations cost a single unit, captions cost far more, and every request costs at least one unit even when it fails. quota resets at midnight pacific.

there is no paid tier. you cannot buy quota. the only route above the default is an audit and quota extension form requiring a compliance review, which community reports put at weeks to months with many applications stalling. spreading load across multiple cloud projects violates the terms and is a common way to lose access entirely.

a hundred uploads a day caps a multi-tenant scheduler at roughly a hundred customers posting once daily. that is a harder wall than a price, because a price you can pay.

there are two tiers, both free. trial allows a thousand requests a day per app; standard raises that to a hundred requests a second per user, with per-category limits published in detail.

the trap is that pins and boards created under trial access are sandbox entities visible only to their creator. your integration will appear to work perfectly — posts succeed, ids come back, no errors — and nobody will ever see the results. it is the same class of problem as tiktok's audit gate and it catches people the same way.

upgrading to standard needs a video recording of your app completing an action through the api, verifying your oauth implementation and data handling, submitted with a privacy policy. reviews take weeks and rejections frequently come without explanation.

the commerce and catalogue apis are genuinely good once you are through, and write limits in trial are only three hundred a day, which makes thorough testing painful.

the community management api covers organisation posts, comments, reactions, follower and page statistics, video analytics and posting on behalf of individual profiles. the data is genuinely valuable for b2b.

getting in is the problem. you need a registered company with a verified linkedin page, a two-tier app review, and a screencast demonstrating each use case. there is a documented gotcha where community management access is greyed out unless you use a brand-new app with no other api products, so people with ads access have to create a throwaway app, get approved, then transfer.

more decisively, the member social permission is closed. linkedin's own faq says it is not accepting access requests due to resource constraints. if your product needs to read an individual member's posts or social activity, linkedin is simply shut to you.

and it breaks you on a timetable: linkedin uses dated api versioning and sunsets old versions on a rolling basis, so recurring migration work is not optional maintenance, it is a permanent line item. for a solo developer or a small team, direct access is realistically unobtainable — which is exactly why every aggregator advertises linkedin support so prominently.

the api itself is reasonable: query creator info, direct post video by file upload or url pull, post photos, check post status. read endpoints allow six hundred requests a minute.

then there is the audit gate, and it deserves to be stated as plainly as tiktok states it: all content posted by unaudited clients is restricted to private viewing mode. unaudited clients are additionally capped at five users posting per twenty-four hours.

the reason this is worse than an ordinary approval process is that nothing fails. your integration works. the api returns success. the posts exist. they are just set to self-only and nobody will ever see them. developers routinely ship this to production before noticing.

lifting it requires an audit verifying terms compliance, with a privacy policy and a demo video of the full upload and oauth flow. tiktok publishes no timeline and no criteria. separately, the research api is now restricted to academic and non-profit use with commercial users explicitly ineligible, so commercial tiktok data access through official channels is effectively gone.

the free tier remains usable for non-commercial work at around a hundred queries a minute per authenticated client, and community libraries like praw still work fine against it.

commercial use is a different world. self-service registration for new commercial oauth access is effectively closed, new tokens need manual approval reported at two to four weeks with no guarantee, and commercial use requires a contract. there is no small paid tier — you go from free and non-commercial straight to enterprise negotiation.

machine-learning use of reddit data is prohibited without a licensing deal, and reddit has instead signed direct ai licensing agreements with major model providers. that is the strategic logic behind all of it.

one note on the numbers you will find elsewhere: a specific enterprise price for fifty million calls circulates widely with identical wording across many sites and traces to no primary source. the arithmetic works out exactly from the per-call rate, which strongly suggests someone calculated it once and everyone copied it. we are not repeating it.

creative kit is a client-side share mechanism, not a server-side api. a human taps a share button and the content opens in snapchat's own camera screen for them to edit and send. there is no way to publish on a user's behalf programmatically.

that means any tool advertising snapchat posting is either using creative kit — which requires the human tap — or doing something unsupported. it is worth checking what an aggregator actually means when it lists snapchat among its networks.

the marketing api is a different and much better story: full programmatic advertising with campaigns, reporting, attribution, audience management and creative upload, open to all advertisers, agencies and developers for years. as paid-social apis go it is among the easiest to get production access to.

so for advertising, snapchat is genuinely accessible. for the organic posting and agent use cases this ranking is written for, it is not gated — it simply does not exist.