X data for research

An X (Twitter) API for research projects, with a free tier and no application to write.

Search public posts by keyword, phrase, account, language and date range, read whole threads, replies and profiles, and save it all as JSON or CSV from Python or R. Weir is read-only, independent of X, and starts free.

Start free Read the quickstart

5,000 free credits every month · no card needed

Is there still a Twitter API for academic research?

Not a free one. X ended the Academic Research track in 2023 and now sells pay-per-use access at $0.005 a post read, with a monthly cap. EU-based researchers can apply for data access under the Digital Services Act. For a study of public posts, a third-party read API such as Weir is the quickest route: search with X's operators, page through results, and export, with 5,000 free credits a month, about 4,700 posts. It is not a full archive: it returns what X's public search serves.

Updated October 2026.

What changed for researchers

Until early 2023, the Academic Research product gave qualifying researchers the full archive for free. In February 2023 X ended free access, and the paid tiers it launched in March started at $100 a month for a small read allowance, with enterprise access reported from $42,000 a month. The Botometer team at Indiana University said the fee would price out most researchers, and the OSoMe tools have run in archival mode since. Today X's pay-per-use pricing charges $0.005 per post and $0.010 per user record, with a cap of 3 million post reads a billing cycle.

Source: Twitter's restrictive API may leave researchers out in the cold, TechCrunch, February 2023 (checked October 2026).

Search like the advanced search page

Keywords, exact phrases, from: and to:, hashtags, language, min likes, date ranges to the millisecond, replies or originals only. The same operators X's own search uses, as JSON fields.

Whole conversations

Threads in order, replies with their cursor, quotes, reposters, edit history and long-form Articles. Profiles carry follower counts, creation dates and the public "account based in" region.

Reproducible and citable

Every post keeps its id, URL and timestamp. The citation generator writes APA 7 and MLA 9 references from a post link, and the search operators guide documents exactly what each filter does.

Free to start, cheap to scale

5,000 credits a month free, no card: about 4,700 posts or 1,600 profiles. After that, credit packs priced per result, with no monthly minimum and no charge for calls that fail.

Collect English posts about a topic from one week, newest first, 20 a page

curl https://api.weirapp.io/v1/x/search \
  -H "Authorization: Bearer $WEIR_KEY" \
  -H "Content-Type: application/json" \
  -d '{"query": "heat wave -filter:nativeretweets", "language": "en", "since": "2026-07-01T00:00:00Z", "until": "2026-07-08T00:00:00Z", "sort": "latest", "limit": 20}'

Every endpoint and its parameters are in the docs.

What it costs

Result	Credits	Per 1,000 results
X post	1	$0.10–$0.20
X profile	2	$0.20–$0.40
X follower / user in a list	1	$0.10–$0.20
Reddit post or comment	3	$0.30–$0.60

Plus one credit per request. 5,000 credits free every month; prepaid packs from $5 that never expire; failed calls are free. Full pricing

Questions

Can I get historical tweets for a study?

Weir returns what X's public search serves for a query and date range, which reaches back years for most topics but is not a complete archive: X's own help says search may not show every post. Describe the collection method and dates in your paper, and keep the ids so the sample can be re-fetched.

Is using Weir for research allowed?

Weir is an independent, read-only service and is not affiliated with X. It returns public content only. You remain responsible for complying with X's terms, your institution's ethics review and the privacy law that applies to you; store only the fields your study needs and avoid identifying private people in what you publish. This is not legal advice.

Does it work from R, Python or Stata?

It is a plain HTTPS JSON API, so anything that can make a web request works: requests in Python, httr2 in R, or curl into a CSV. The Python guide on the blog has a tested collection script with paging and CSV export.

What about Reddit data?

The same key reads public Reddit posts, comments and user histories, which many studies pair with X. Reddit data is 3 credits a result and is subject to Reddit's policies for researchers.

Is there a discount for students or universities?

Every account gets the same free 5,000 credits each month, which covers a pilot study. For larger collections, write to support with the study size and we will quote a pack.

Guides

Start free