Twitter/X API & scraper for academic research

Search the full archive back to 2006, rehydrate post IDs 100 at a time, pull timelines, replies and followers. Academic access is tailored to your study: from requests granted to your project to deep discounts.

Apply for academic access Start with 100 free requests

No credit card, and they never expire.

For my I need about

About 2 hoursto collect at 20 requests per second

$15,000on the official X API

from $210on Sorsa API, before academic access

Academic access goes further: tailored to your study, from granted requests to deep discounts.

Apply with this plan

Academic access, built around your study

Requests granted for your study

We add the requests your project needs to your Sorsa API account.

A deep discount on a plan

For long or large projects: any plan at an academic price.

What every approved project gets

Every endpoint
20 requests per second
The same docs and API as everyone
Fast answers from the team

Who can apply

PhD students
Master’s students
Faculty
Postdocs
Research staff
NGO and newsroom researchers

Non-commercial research in any discipline, applied for from an institutional email.

Apply in about two minutes. Meanwhile, your 100 free requests work on every endpoint.

Apply for academic access

Every method your study needs

One API key, the same endpoints for 2006 and for this morning.

Archive search
Keyword corpora, event windows, long time series.
"heat wave" lang:en since:2022-07-01 until:2022-08-01 -filter:nativeretweets
About 20 posts a request, back to 2006
Posts by ID
Rehydrate a shared dataset or replicate a published corpus.
POST /tweet-info-bulk {"tweet_links": [100 IDs]}
100 posts a request, each with its author
Account timelines
Panels of politicians, journalists or organisations.
from: + dates, or /user-tweets for an account’s whole history
About 20 posts a request, no 3,200-post cap
Conversations
Reply trees, quote chains, who retweeted.
/comments, /quotes, /retweeters
About 20 a request
Networks
Followers and following of any public account.
/followers?username=…
200 profiles a page
Profiles and account context
Profiles 100 at a time; country, username changes and verification from About this account.
/info-batch, /about
100 profiles a request
Lists, Communities, Spaces
One X List watches up to 5,000 accounts; community posts and members; Space metadata.
/list-tweets?list_id=…
About 20 posts a request

Search operators, pagination and page sizes are in the docs: Historical data, Search operators, Posts by ID.

Every post arrives with its variables

Coastal Climate Lab@coastal_climate_lab

Replying to @heat_records_archive

Our open dataset is out: 1.2 million posts about the 2022 European heat waves, with the post IDs and the code to rehydrate them. #heatwave

Posts a day, June to August 2022

10:42 AM, Aug 18, 2023

96.4K Views211 Bookmarks

42 Replies318 Reposts27 Quotes1.2K Likes

userusername, followers_count, created_at, verified, location

conversation_id_strin_reply_to_tweet_idIts place in a thread

full_textlangEvery word, and its language

entitiesMedia and links

created_atPosted, to the second

view_countbookmark_countViews and bookmarks

likes_countretweet_countreply_countquote_countEngagement, as it stands

Old posts come with the same fields as today’s.

Plan your design around these

Deleted posts and protected accounts aren’t returned. A rehydrated list comes back smaller than the original.
Counts are today’s. Likes, views and followers are as they stand when you collect; for counts over time, collect on a schedule and keep each snapshot.
Profiles are current. A post from 2014 comes with its author’s profile as it is now.
You poll, nothing streams. New posts arrive when you ask; one request reads an X List of up to 5,000 accounts.
Not available: who liked a post, a user’s likes or bookmarks, DMs, past follower lists.

Try your date range on the free requests before you plan the full collection.

From an ID list to a dataset

rehydrate.py

import csv, json, time, requests

URL = "https://api.sorsa.io/v3/tweet-info-bulk"
KEY = "YOUR_API_KEY"
ids = [row[0] for row in csv.reader(open("tweet_ids.csv"))]

with open("tweets.ndjson", "w") as out:
    for i in range(0, len(ids), 100):
        while True:
            r = requests.post(URL, headers={"ApiKey": KEY},
                              json={"tweet_links": ids[i:i + 100]})
            if r.status_code != 429:
                break
            time.sleep(1)  # 20 requests per second per key
        r.raise_for_status()
        for tweet in r.json()["tweets"]:
            out.write(json.dumps(tweet) + "\n")

rehydrate.R

library(httr2)
library(jsonlite)

ids <- readLines("tweet_ids.txt")
batches <- split(ids, ceiling(seq_along(ids) / 100))

fetch <- function(batch) {
  resp <- request("https://api.sorsa.io/v3/tweet-info-bulk") |>
    req_headers(ApiKey = Sys.getenv("SORSA_API_KEY")) |>
    req_body_json(list(tweet_links = as.list(batch))) |>
    req_retry(max_tries = 5) |>
    req_perform()
  resp_body_json(resp)$tweets
}

tweets <- unlist(lapply(batches, fetch), recursive = FALSE)
write_json(tweets, "tweets.json", auto_unbox = TRUE)

A million IDs take about eight minutes at 20 requests per second.
Failed requests are never charged.
Writing it with an AI assistant? Give Claude Code, Cursor or Codex our docs file, or connect the MCP server. Build with AI

For your ethics board and your methods section

What reviewers ask

What data?: Public posts and profiles only. No protected accounts, no DMs.
Any interaction?: None. Sorsa API is read-only: it never posts, likes, follows or messages.
Whose account?: No X account or login is involved, yours or anyone’s.
What does the provider keep?: Which requests you made, for a year; the full responses, for 30 days.
Who knows about the project?: X never learns who is asking: your key, account and identity stay with us.
Where does your copy live?: Wherever your data management plan says. It’s yours.
Paperwork?: On request, we write a short letter about the data source for your ethics board.

The methods paragraph

2.1 Data collection

Posts were collected through Sorsa API (https://api.sorsa.io), an independent API for public X (Twitter) data, using full-archive search for 1 January to 31 March 2024.

Questions researchers ask

PhD and master’s students, faculty, postdocs and research staff at universities and research institutes, for non-commercial research in any discipline. Apply from your institutional email.

Access tailored to your study: from the requests it needs, granted to your account, to a deep discount on a plan. Every endpoint, 20 requests per second, the same docs and support as every customer.

A person on our team reads your application and answers by email. If something is unclear we ask; then we agree the volume and the period with you and add the access to your Sorsa API account.

Yes. Every account gets 100 free requests: no card, they never expire, and they work on every endpoint. That’s up to 10,000 posts by ID or 20,000 follower profiles, enough to pilot your queries.

Apply with the email you use and add a link to your faculty, lab, ORCID or Google Scholar page, so we can see your affiliation.

To 2006. Full-archive search takes X’s own operators: from:, to:, since:, until:, lang:, exact phrases, hashtags and engagement filters. Account timelines go back to the first post, without a 3,200-post cap.

Yes: 100 IDs per request, each post with its author’s full profile. A million IDs take 10,000 requests, about eight minutes at 20 requests per second. Posts deleted or made private since the original collection don’t come back, so expect fewer posts than IDs.

Deleted posts, protected accounts, DMs, lists of who liked a post, a user’s likes and bookmarks, and follower lists as they were in the past. Counts and profiles are as they stand when you collect. Nothing streams: you poll, and one request reads an X List of up to 5,000 accounts.

That the data is public posts and profiles; Sorsa API is read-only and never posts, likes, follows or messages; no X account or login is involved; and you don’t contact the people in the data. We keep a log of your requests for a year and the full responses for 30 days. On request, we write a short letter about the data source.

Share post IDs and your collection code: anyone can rehydrate them and check your work. How much more you share is up to your ethics approval and data management plan.

In your methods section: “Data were collected through Sorsa API (https://api.sorsa.io), an independent API for public X (Twitter) data.” The BibTeX is in section 5 on this page.

Yes. Your account holds several API keys: give each person their own and see each key’s usage in the dashboard.

Yes. Say so in the application and we’ll arrange an invoice for your project.

X closed its Academic Research track in 2023. Its pay-per-use API bills $0.005 per post and $0.010 per profile; its EU route under the Digital Services Act covers only research on systemic risks in the EU, through a separate application. Sorsa API works for any discipline and topic, with one key and no X developer account.

TwitterAPI.io refunds half of what a research project spends, up to $200, after you pay. Sorsa API tailors access to your study before you collect, from granted requests to deep discounts.

Then a plan fits: posts from $0.014 per 1,000, every endpoint on every plan. For bigger volumes, talk to sales.

Twitter/X API & scraper for academic research

Academic access tailored to your study, from granted requests to deep discounts.

Apply for academic access Start with 100 free requests

Academic access, built around your study

Requests granted for your studyWe add the requests your project needs to your Sorsa API account.
A deep discount on a planFor long or large projects: any plan at an academic price.

Apply for academic access

Every method your study needs

One API key, the same endpoints for 2006 and for this morning.

Every post arrives with its variables

Coastal Climate Lab@coastal_climate_lab

user

Replying to @heat_records_archive

in_reply_to_tweet_id

Our open dataset is out: 1.2 million posts about the 2022 European heat waves, with the post IDs and the code to rehydrate them. #heatwave

full_text

10:42 AM, Aug 18, 2023

created_at

42 Replies318 Reposts27 Quotes1.2K Likes

likes_count …

Old posts come with the same fields as today’s.

From an ID list to a dataset

rehydrate.py

import csv, json, time, requests

URL = "https://api.sorsa.io/v3/tweet-info-bulk"
KEY = "YOUR_API_KEY"
ids = [row[0] for row in csv.reader(open("tweet_ids.csv"))]

with open("tweets.ndjson", "w") as out:
    for i in range(0, len(ids), 100):
        while True:
            r = requests.post(URL, headers={"ApiKey": KEY},
                              json={"tweet_links": ids[i:i + 100]})
            if r.status_code != 429:
                break
            time.sleep(1)  # 20 requests per second per key
        r.raise_for_status()
        for tweet in r.json()["tweets"]:
            out.write(json.dumps(tweet) + "\n")

rehydrate.R

library(httr2)
library(jsonlite)

ids <- readLines("tweet_ids.txt")
batches <- split(ids, ceiling(seq_along(ids) / 100))

fetch <- function(batch) {
  resp <- request("https://api.sorsa.io/v3/tweet-info-bulk") |>
    req_headers(ApiKey = Sys.getenv("SORSA_API_KEY")) |>
    req_body_json(list(tweet_links = as.list(batch))) |>
    req_retry(max_tries = 5) |>
    req_perform()
  resp_body_json(resp)$tweets
}

tweets <- unlist(lapply(batches, fetch), recursive = FALSE)
write_json(tweets, "tweets.json", auto_unbox = TRUE)

A million IDs take about eight minutes at 20 requests per second.
Failed requests are never charged.
Writing it with an AI assistant? Build with AI

For your ethics board and your methods section

What data?

Public posts and profiles only. No protected accounts, no DMs.

Any interaction?

None. Sorsa API is read-only: it never posts, likes, follows or messages.

Whose account?

No X account or login is involved, yours or anyone’s.

What does the provider keep?

Which requests you made, for a year; the full responses, for 30 days.

Who knows about the project?

X never learns who is asking: your key, account and identity stay with us.

Where does your copy live?

Wherever your data management plan says. It’s yours.

Paperwork?

On request, we write a short letter about the data source for your ethics board.

Apply for academic access

About two minutes. A person on our team reads every one.

1Send
2We read it
3We agree
4You collect

Start with 100 free requests

Prefer email? Write to contacts@sorsa.io from your institutional address.

Twitter/X API & scraper for academic research

1Academic access, built around your study

2Every method your study needs

3Every post arrives with its variables

4From an ID list to a dataset

5For your ethics board and your methods section

Apply for academic access

Questions researchers ask

Who can apply?

What does academic access include?

What happens after I apply?

Can I start before I hear back?

I don’t have an institutional email.

How far back does the data go?

Can I rehydrate a dataset of post IDs?

What isn’t in the data?

What should I tell my ethics board?

Can I share my dataset?

How do I cite Sorsa API?

Can my co-authors use the same access?

Can you invoice my university?

What does X offer researchers now?

How does this compare with other academic offers?

My project is commercial.

Read next

Twitter/X API & scraper for academic research

Academic access, built around your study

Every method your study needs

Every post arrives with its variables

From an ID list to a dataset

For your ethics board and your methods section

Apply for academic access

Questions researchers ask

Read next

Academic access, built around your study

Every method your study needs

Every post arrives with its variables

From an ID list to a dataset

For your ethics board and your methods section