WEIR
Get a key
Theme

Twitter Data for Academic Research: 2026 Guide

Twitter data for academic research after the free API ended: X's new route for EU studies, what a 10,000-post dataset costs, and an ethics checklist.


Published · Updated · 8 min read

Cover image: Research data on a grant budget

Twitter data for academic research now costs money: X's API charges per post, and the free academic track ended in 2023.

This guide is for students, faculty and research staff who need X (Twitter) posts for a study. You get what changed, what a 10,000-post dataset costs on each route, how to keep it reproducible, and a terms checklist. We run Weir, one of the APIs below, so weigh our view knowing that.

A month of English posts on one topic, by API:

curl --request POST \
  --url https://api.weirapp.io/v1/x/search \
  --header "Authorization: Bearer $WEIR_API_KEY" \
  --header 'Content-Type: application/json' \
  --data '{"query":"heatwave","language":"en","since":"2026-07-01T00:00:00Z","until":"2026-08-01T00:00:00Z","limit":20}'
$50for 10,000 posts on X's API
10Mposts a month the old free track allowed
3Mpost reads a month, X's pay-per-use cap

What happened to Twitter's academic API?

Twitter gave academics free, deep access for about two years, then priced it like everyone else's:

  • January 2021: Twitter launched the Academic Research track. TechCrunch reported free full-archive search and a cap of 10 million tweets a month.
  • February 2023: Twitter said it would end free API access. TechCrunch reported that a $100 basic tier would replace legacy levels, Academic Research included.
  • February 2026: X launched pay-per-use. Its @XDevelopers post of 6 February reads "Officially launching X API Pay-Per-Use".

Most search results for "Twitter academic API" still describe the 2023 shutdown. That's history now. The useful question is what works today.

The X API for researchers in 2026, and two other routes

You have three routes. Which one fits depends on your topic, your budget and your ethics board.

The official X API, pay-per-use

X's pricing page lists $0.005 per post read and $0.010 per user read, as of October 2026. It names no academic or student discount.

Full-archive search reaches back to March 2006, per X's search docs, at up to 500 posts per request.

Pay-per-use accounts read at most 3 million posts a month.

Good

  • X's own data, terms and support
  • Full archive, 500 posts per request

Watch out

  • $0.005 per post, with no academic price
  • Bars training foundation models on the data

Free access for research on EU risks

The EU's Digital Services Act gives researchers a legal route to platform data. X's help page on it names two doors.

Requests under Article 40(4) go through the European Commission's DSA Data Access Portal. Requests under Article 40(12), for public data, go through X's own DSA Researcher Application Form.

The scope is narrow: X's Developer Agreement limits this access to research on systemic risks in the European Union.

On 16 July 2026, the Commission accepted X's plan to fix this. X promised faster screening, free data, and terms that stop banning eligible researchers from scraping. It has six months to deliver, so check the current state before you apply.

Third-party APIs

Services like Weir and TwitterAPI.io read public posts for you and bill per result. Neither is X, and neither is licensed by X, so access can change.

Good

  • Cents per thousand posts, no X developer account
  • Plain HTTP or an AI assistant via MCP

Watch out

  • Not X's official data or terms
  • Read-only on Weir; no post counts or compliance feeds

How much does Twitter data for a study cost?

Take a typical project: 10,000 posts on one topic, collected once.

Collecting 10,000 X posts
Official X API$50
Weir, $5 packabout $2.10
TwitterAPI.ioabout $1.50

On Weir, a search call costs 1 credit plus 1 per post. X serves about 20 posts per call, so 10,000 posts take 500 calls and 10,500 credits.

The $5 pack holds 25,000 credits that never expire.

The 5,000 free credits each month cover about 4,700 posts, which is enough for a pilot.

TwitterAPI.io is cheaper per post than we are. Its pricing page lists $0.15 per 1,000 tweets.

Run your own numbers in our X API pricing guide.

How to download Twitter data for research you can repeat

A dataset nobody can rebuild is hard to defend in peer review. Four habits help:

  1. Save the exact query and the date you ran it. Search results change as posts are deleted and accounts go private.
  2. Store post IDs as strings. They exceed what a float can hold, and Excel rounds them.
  3. Record when you read the counts. Likes and views keep moving after you collect.
  4. Share IDs, not text, so others fetch the posts fresh.

Counts really do drift. We fetched X's first public post twice on 21 October 2026, and its likes went from 311,641 to 311,642.

To rebuild a dataset from stored IDs, look them up 100 at a time:

curl --get https://api.weirapp.io/v1/x/posts \
  --header "Authorization: Bearer $WEIR_API_KEY" \
  --data-urlencode "ids=20,2019881223666233717,1"

We ran that lookup on 21 October 2026. Two posts came back, and the missing one was listed, not billed:

{
  "data": [
    { "id": "20", "text": "just setting up my twttr", "createdAt": "2006-03-21T20:50:14.000Z" },
    {
      "id": "2019881223666233717",
      "text": "Officially launching X API Pay-Per-Use ...",
      "createdAt": "2026-02-06T21:09:45.000Z"
    }
  ],
  "meta": { "notFound": ["1"] }
}

It cost 3 credits.

A post that was deleted after you collected it lands in notFound the same way, so a rebuilt dataset respects deletions.

If you collect through X's own API, check its Developer Agreement first. It counts post IDs as X content, and it bars redistributing that content without X's permission.

For collection code, see how to get tweets with Python. For one account's full history, see how to download all tweets from a user. Our advanced search operators guide helps you write precise, repeatable queries.

An ethics and terms checklist

Go through this before you collect, not after. This isn't legal advice:

  • Ask your ethics board early. Public posts still come from people, and your IRB or ethics committee decides what review you need.
  • Read X's terms. X's Terms of Service forbid scraping without X's written consent, and the Developer Agreement adds rules for API users.
  • Check the privacy law that applies, such as the EU's GDPR, both to you and to the people in your data.
  • Don't train AI models on the posts. X's Developer Agreement bars using X content to train or fine-tune a foundation or frontier model.
  • Collect the minimum. Drop fields you don't need, and pseudonymise handles in what you publish.
  • Never single people out. Weir's acceptable use policy forbids building profiles of specific people without a lawful basis.

When not to use Weir

  • Your study is about systemic risks in the EU. Apply for DSA access, where X has promised free data.
  • Your funder or journal requires official data. Use X's API and its terms.
  • Volume decides everything. TwitterAPI.io lists a lower price per tweet, and it can also post and stream.
  • You need post counts or compliance feeds. Those come from X's own API.

FAQ

Is the Twitter academic API still available?

No. Twitter ended the free Academic Research track in 2023 and replaced it with paid tiers. Since February 2026, X sells pay-per-use access with no academic discount. Research on EU systemic risks can apply under the Digital Services Act instead.

How much does Twitter data cost for a research project?

About $50 per 10,000 posts on X's API, at $0.005 each as of October 2026. Third-party APIs charge less: about $2.10 on Weir's smallest pack and about $1.50 on TwitterAPI.io.

Can students get free Twitter API access?

X's pricing page lists no free tier and no student discount. Weir gives everyone 5,000 free credits a month, about 4,700 posts, with no card. EU risk research may qualify for free data through X's DSA process.

Can I share a Twitter dataset with other researchers?

Share post IDs and your exact queries, not full text. Others can then rebuild the dataset, and deleted posts drop out. If you used X's API, check its Developer Agreement first, since it limits redistribution.

Sources

Keep reading