X API Pricing 2026: Twitter API Pricing and Cost Per Post

What does the X API cost in 2026?

It costs $0.015 to publish a post, and $0.200 to publish a post that contains a link. There is no monthly plan to buy at the self-serve level any more. X's developer product is now called the pay-per-use X API in its own documentation, it is credit based, you buy credits up front and they are deducted as you call the API. X's own material is blunt about the model: no contracts, no minimum spend, stop whenever you like.

The rates below were read directly off the pricing endpoint X's own documentation site calls to build its cost estimator, on August 31, 2026. They are the live numbers, not a roundup's recollection of them.

One detail worth having if you are budgeting: buying X API credits earns you xAI API credits back, up to 20% of your spend. If you were already going to spend money on Grok, that materially changes the real cost of the X side, and almost no pricing article mentions it.

Why does a post with a link cost $0.200 instead of $0.015?

Because X prices link-bearing posts as a separate request type. In the rate card the ordinary publish call is PostCreate at $0.015, and a post carrying a URL is ContentCreateWithUrl at $0.200. Same tweet, same character count, same account. Add a link and it costs 13 times more to publish.

X does not explain the reasoning, and it would be dishonest to pretend to know it. What is observable is the effect. Link posts are how marketers push traffic off X to a blog, a product page or a signup form, so the surcharge falls almost entirely on commercial posting rather than on conversation. A person tweeting their thoughts pays the cheap rate. A company doing content distribution pays the expensive one.

There is one exception in X's own note on the pricing page and it is easy to miss: summoned replies stay at the standard price. A reply the API makes because it was mentioned is not hit with the URL surcharge in the same way. If you are building a support or reply bot rather than a publishing pipeline, your economics look completely different from the numbers most people quote.

What happened to the $200 Basic tier?

It is gone from self-serve. X's documentation still carries Free, Basic and Pro columns in its rate limit tables, which is why so many current articles describe a tier structure you can no longer buy, and the navigation now leads with pay-per-use instead. Enterprise still exists as a real, separate thing: several endpoints are marked as requiring an Enterprise plan and explicitly not available on the self-serve pay-per-use tier.

Reporting through 2026 has it that legacy Basic at $200 a month was retired and its subscribers moved onto pay-per-use, and that Pro at $5,000 a month closed to new signups. We have not verified those two figures at the source and are flagging them as reported rather than confirmed, because the pages that carried them are no longer public. What we can confirm first-hand is the current state: the self-serve product is credit based and metered per request, and the tier names survive only in rate limit documentation.

The practical consequence for anyone comparing build against buy is that the old reasoning has inverted. Under $200 a month flat, heavy posting was cheap and light posting was wasteful. Under pay-per-use, light posting is nearly free and heavy link posting is where the bill appears.

What does automated posting actually cost per month?

Run the two rates against realistic cadences and the spread is dramatic. This is the same arithmetic on the same source figures, once for plain text and once for the link-bearing posts most brands actually send.

A company posting ten link-bearing tweets a day spends about $60 a month on API calls alone, before a line of code has been written or maintained. The same volume without links is $4.50. If your posting mix is roughly half links, budget somewhere near $32 for that cadence and re-check it whenever the mix changes, because the mix, not the volume, is what moves this number.

Can you schedule posts through the X API?

No, and this surprises people who assume scheduling is a platform feature that tools simply expose. Search X's API documentation for the word schedule and the only matches concern audio Spaces. There is no scheduling parameter on post creation at all. The API publishes now or not at all.

That single gap explains the entire third-party scheduling market and one of its most common failure modes. When you set a post for Tuesday at 9am in any tool, nothing is stored at X. The post sits in the vendor's database, and at 9am their system wakes up and makes an ordinary live publish call. So the vendor being online at that minute, not X being online, is what decides whether your post goes out. We work through what that means when choosing a tool on our Twitter scheduler comparison.

If you build this yourself, you own that reliability problem outright. You need a queue, a retry policy, token refresh handling for connections that silently expire after a password change, and something watching the whole thing, because a posting job that dies quietly is worse than one that never existed. Teams that run their own integration usually end up watching the endpoint on a short interval rather than finding out from a client that nothing published for three days.

Should you build on the X API or buy a scheduler?

Buy, unless posting to X is the product you are selling. The arithmetic is not close for ordinary marketing use.

Take the ten-link-posts-a-day company again. Raw API cost is roughly $60 a month. Buffer charges $5 a month for that channel, Publer $5 for one account, Zoho Social $10 a month billed annually for a brand covering multiple channels. The flat-rate vendor is cheaper than the API rate card before you count the engineering, and you also get a calendar, a mobile app, threads, analytics and someone else's on-call rota.

Building makes sense in a narrower set of cases: you are shipping a product that posts on behalf of your own customers, you need posting behaviour no vendor exposes, or your volume and negotiating position justify an Enterprise conversation with X directly. If you are a marketing team wanting your posts to go out at 9am, the API rate card is an argument for buying, not building.

The one thing no scheduler solves is having something worth posting ten times a day. That is a production problem rather than a distribution one, and it is where our own tool sits: turning one source asset into channel-native posts you load into whichever scheduler you already pay for. The Twitter thread generator covers the X side on its own, and the social media manager tools comparison covers the wider stack.

What do reads and analytics cost?

Reads are cheap individually and dangerous in bulk. Reading a post costs $0.005, looking up a user costs $0.010, and pulling trends costs $0.010. Reading your own account's data is billed at the much lower OwnedRead rate of $0.001, which matters more than it looks: an analytics dashboard for your own account is roughly a fifth the price of one that reads the wider timeline.

The trap is scale. At $0.005 a post, pulling 100,000 posts for a listening or research project is $500, and it is very easy to write a loop that does that without meaning to. Anyone building on this should set a credit ceiling in the Developer Console and watch the live usage tracking rather than reconciling at the end of the month.

For most marketing teams this is another argument for buying rather than building. The analytics you get inside a $5 to $35 a month scheduler are already paid for out of the vendor's own agreement with X, and reproducing them on the self-serve rate card would cost more than the subscription.

How to keep this from surprising you

Three habits cover most of it. First, price your posting mix rather than your posting volume, because the link surcharge is the variable that actually moves the bill. Second, treat the tier names you find in older articles with suspicion; the self-serve product is credit based now and several widely cited monthly figures describe plans you cannot buy. Third, if you are choosing between building and subscribing, do the comparison at your real cadence with the link split included, because at ordinary marketing volumes a flat-rate tool wins by a wide margin.

All the rates on this page came from X's own pricing endpoint and documentation on August 31, 2026. X has changed this pricing more than once, so check the figures against the Developer Console before you commit a budget to them.