Since X moved to pay-per-usage credits, a cluster of third-party read APIs has grown up underneath it. They all promise the same thing — public X data, far cheaper — and their pricing pages are close to useless for comparison because each one quotes a different unit. This is an attempt at an honest like-for-like, verified 15 September 2026.
Disclosure up front: Probe Uranus is one of the options listed. The numbers below are published list prices with links to their sources, and the section on where we lose is not decoration — read it before you buy.
Published prices
| Service | Model | Per 1,000 posts | Per 1,000 profiles | Try before paying |
|---|---|---|---|---|
| Probe Uranus | Flat monthly | ~$0.05 at full use | ~$0.05 at full use | Free trial, card required |
| Official X API | Metered per call | $5.00 | $10.00 | Nothing — credits are bought up front |
| TwitterAPI.io | Metered per call | $0.15 | $0.18 | $0.10 in free credits, no card |
Other services in this space — SocialData, GetXAPI, Sorsa, TwitterAPIs — advertise rates roughly between $0.02 and $0.20 per 1,000 tweets. They are left out of the table because we have not verified their current list prices directly, and a table of half-checked numbers is worse than a short one.
The unit problem
Before comparing any two of these, work out what each one counts:
- Per resource. The official API bills every post, user, and like it returns. One paginated search can cost hundreds of units.
- Per call. Most third-party APIs bill the request, so a call returning 100 tweets costs the same as one returning 1. This rewards large page sizes heavily.
- Flat monthly. A fixed bill against per-minute, daily, and monthly caps. The effective per-1,000 rate depends entirely on how much of the allowance you use — quoting it as a headline number, as we do, is only honest with the "at full use" caveat attached.
A flat plan and a metered plan cross over at a specific volume. Ours is on the comparison page, computed from the live price rather than written into prose.
What the pricing pages leave out
Reliability is the actual product
Every service here reads X through infrastructure X does not officially sanction. That means all of them degrade in the same way at the same time when X changes something. The question is not whether a provider breaks, but how fast they notice and fix it. Ask about status pages and incident history before you ask about price; a $0.02 rate on an endpoint that returns empty arrays for two days is not cheap.
Coverage is uneven and usually overstated
"Full coverage" typically means profiles, timelines, and search. The long tail — Spaces, communities, broadcasts, quote trees, thread context, follow relationships, list members — varies a lot, and some of it is thin or empty upstream no matter who is asking. Test the two or three endpoints your product actually needs before committing; the aggregate endpoint count on a landing page tells you nothing.
Nobody here can write
If a third-party service offers posting, liking, or following, it is doing so with credentials somewhere in the loop. Understand whose account carries the risk. Read-only services, including this one, cannot put your X account in jeopardy because they never touch it.
Terms and data handling
Check what the provider stores, for how long, and whether they retain the right to reuse what you query. For a side project this is noise; for anything with users attached it is the clause that matters most, and it is rarely on the pricing page.
A short decision procedure
- Need writes, DMs, ads, or streaming? Official API. Nothing else qualifies. See what that costs in 2026.
- Spiky or low volume? Metered, pay-as-you-go. A quiet month should cost near nothing, and a flat plan will not give you that.
- Steady volume above the break-even? A flat plan removes the meter anxiety and makes the line item predictable.
- Evaluating right now? Prefer whoever lets you make a real call today. Free credits without a card are the cheapest way to find out that an endpoint does not return what the docs imply.
Where Probe Uranus loses
So you do not have to find out after paying:
- Low volume. Below the break-even on /compare, metered credits are cheaper.
- No card-free trial. Ours needs a card up front, even though nothing is charged until the trial ends. Some competitors do not. If that is a blocker, it is a fair one.
- Reads only. No posting, no DMs, no streaming, no historical archive.
- A monthly ceiling. The public plan caps out; very large workloads need a conversation rather than a checkout.
What we do have is a flat bill, quotas you can query at any time, a public OpenAPI document and llms-full.txt for agents, and an engager endpoint that draws seeded random winners from a tweet. You can try that last one without an account on the giveaway picker.
All prices verified 15 September 2026 against the vendors' own pages and subject to change. Check before you buy.