# How to scrape Twitter

How to scrape Twitter comes down to five routes, each with a different price, reliability and legal risk. We checked all five using X's own docs, the GitHub API, the Apify Store API and RapidAPI's listing data.

> We make FATSEARCH. The numbers below are our own runs, dated. Where another reader is the better pick, we say so.

## X scraper: the official pay-per-use API

X switched to pay-per-use pricing on Feb 6, 2026 ([dev community post](https://devcommunity.x.com/t/announcing-the-launch-of-x-api-pay-per-use-pricing/256476)). A post read costs $0.005 ([pricing page](https://docs.x.com/x-api/getting-started/pricing)), capped at 3 million Post reads per monthly billing cycle ([same page](https://docs.x.com/x-api/getting-started/pricing)). That cap contradicts the marketing page on developer.x.com, which says "No monthly caps".

This is the only route where you have a contract with X. It is also the most expensive per tweet.

## Scrape x.com: open-source GitHub tools

We checked the top GitHub repos for X/Twitter scraping by stars (via the GitHub API). The decay is real. The two most-starred are archived: twint (16,405 stars, last commit 2021-03-02) and nitter (14,469 stars). snscrape (5,464 stars) last committed on 2023-06-22, and got 35 breakage-titled issues in 2023. Stars in archived or dormant repos account for 80% of the total (43,201 of 53,992).

The breakage is tied to a specific event: X shut down guest-token access. Every repo still alive requires logged-in accounts and carries open issues about anti-bot measures: twikit reports a "palm-liveness scan on login" and "ClientTransaction broken"; twscrape reports "Cloudflare is blocking all programmatic logins"; the-convocation reports "Bad Guest Token".

Free code, but you maintain it yourself, and X's terms apply to your accounts.

## Apify twitter scraper

[Apify](https://apify.com/) is a scraping platform with a marketplace of "actors". The top Twitter actor by users is Tweet Scraper V2 by apidojo (110,690 users, $0.0004 a tweet). The cheapest per tweet is kaitoeasyapi at $0.00025. Twitter Scraper by danek charges $0.0003 a result.

Reliability by run status: Tweet Scraper V2 had 0.18% failed or timed-out runs across 4,337,696 runs over the past month (Apify Store API). That measures whether the run finished, not whether the data was correct.

Apify actors churn too. The original quacker/twitter-scraper (35,577 users) is now marked deprecated.

## RapidAPI twitter api

[RapidAPI](https://rapidapi.com/) is a marketplace where third-party developers list APIs (see our [RapidAPI scrapers guide](/guides/rapidapi-scrapers/) for how the marketplace works). The top Twitter listing by subscriptions is twitter-api45 by Alexander Vikhorev (30,478 subscriptions). It carries 42 billing plans, including private custom ones.

RapidAPI bills per request, not per tweet. The per-request pricing can't be directly compared to per-tweet pricing without knowing the page size.

The listings say "Unofficial" themselves. The API terms are solely between the buyer and the listing provider, not RapidAPI ([RapidAPI Terms](https://rapidapi.com/), updated May 13, 2026). NO REFUNDS WILL BE ISSUED ([same terms](https://rapidapi.com/)).

In our Google checks for generic scraping searches, RapidAPI didn't show up at all. It ranks only when the search already says "rapidapi".

## Unofficial twitter api: standalone sellers

A newer category: dedicated services that sell X data without a marketplace in between.

- [TwitterAPI.io](https://twitterapi.io/pricing): $0.15 per 1K tweets, billed per result, minimum $0.00015 per request. It ranks in the top results for "twitter api alternative".
- [TwttrAPI](https://twttrapi.com/): $59/mo for 2,000,000 requests (Pro plan).
- Xquik: $0.00015 per tweet, sold through a GitHub README that is the top Google result for "twitter scraper". The repo is a README for a paid API, not an open-source tool.

## Where paid routes break too

The thesis that "pay and it works" has limits. Apify's original Twitter scraper (quacker, 35,577 users) is deprecated. On RapidAPI, dead listings stay listed and sellable: Socialgrep shows a 0% success rate and still offers a $100/mo plan. Paying doesn't buy stability. It hands the upkeep to someone who might also stop.

## Legal: what the record shows

This is not legal advice. These are the facts we found, each with its source.

**X's Terms of Service** (effective Oct 9, 2026): "crawling or scraping the Services in any form, for any purpose without our prior written consent is expressly prohibited" ([x.com/en/tos](https://x.com/en/tos)). Liquidated damages: "$15,000 USD per 1,000,000 posts" accessed in any 24-hour period ([x.com/en/tos](https://x.com/en/tos)). That clause was added in the ToS effective November 15, 2024 ([Knight First Amendment Institute](https://knightcolumbia.org/content/knight-institute-says-xs-new-terms-of-service-will-stifle-independent-research)).

**X Corp. v. Bright Data** (N.D. Cal., May 9, 2024): Judge Alsup dismissed X's claims, in part as preempted by the Copyright Act: "risks the possible creation of information monopolies that would disserve the public interest" ([Courthouse News](https://www.courthousenews.com/judge-tosses-xs-contract-claims-against-data-scraping-company/)).

**Meta v. Bright Data** (N.D. Cal.): summary judgment for Bright Data on "logged off" scraping ([Quinn Emanuel](https://www.quinnemanuel.com/the-firm/news-events/client-alert-meta-v-bright-data-significant-decision-for-web-scraping-industry/)). Meta has dropped its lawsuit ([TechCrunch](https://techcrunch.com/2024/02/26/meta-drops-lawsuit-against-web-scraping-firm-bright-data-that-sold-millions-of-instagram-records/), via SERP snippet).

**hiQ v. LinkedIn** (December 2022): $500,000 consent judgment against hiQ, plus a permanent injunction (reported by [Proskauer](https://www.proskauer.com/blog/hiq-and-linkedin-reach-proposed-settlement-in-landmark-scraping-case), via SERP snippet).

**GDPR (EU)**: "Met scraping worden bijna altijd persoonsgegevens verzameld" (scraping almost always collects personal data), Dutch DPA, 02 april 2025 ([autoriteitpersoonsgegevens.nl](https://www.autoriteitpersoonsgegevens.nl/documenten/handreiking-scraping-door-particulieren-en-private-organisaties)).

## Where FATSEARCH fits

FATSEARCH is not a scraping API. It reads one public page at a time for an AI agent, through the user's own provider key (Oxylabs or Bright Data). If your agent needs to read a single tweet or profile page as part of answering a question, FATSEARCH does that: it fetches the page, turns it into a short Markdown outline, and the agent reads only the section it needs.

It does not do bulk data, search by hashtag, or pull follower lists. For those jobs, the routes above are the right tool. For reading one page in context, see [how to set it up](/guides/web-search-mcp-for-claude-code/) or [what it returns vs other readers](/compare/).

Be clear on its limits. Our [benchmarks](/docs/benchmarks/) include Reddit, TikTok, Instagram and YouTube pages, and FATSEARCH lost some of them: Instagram posts came back as empty shells, and some TikTok renders timed out. We have not tested FATSEARCH on x.com pages.

For background on how RapidAPI works as a marketplace and how to vet a listing: [RapidAPI scrapers](/guides/rapidapi-scrapers/).

## Questions

### Is Twitter scraping legal?

X's terms say scraping without prior written consent is expressly prohibited, and set liquidated damages of $15,000 USD per 1,000,000 posts accessed in any 24-hour period (added November 15, 2024). In X Corp. v. Bright Data (May 9, 2024) the judge dismissed X's contract claims. In the EU, scraped posts are personal data under GDPR (Dutch DPA, 02 april 2025). This is not legal advice: what the record shows is that terms, copyright and data-protection law each apply.

### Can you still scrape data from Twitter?

Technically yes, but the free route got much harder. When X shut off guest access, most open-source scrapers that relied on it broke. The repos still alive need logged-in accounts and carry open issues about anti-bot measures: Cloudflare blocks, palm-liveness scans, bad guest tokens. The paid services (Apify actors, RapidAPI listings, standalone APIs) handle authentication on their side.

### Which Twitter scraper is the best?

It depends on the job. If you want a contract with X, its own pay-per-use API at $0.005 per post. For volume at the lowest per-tweet price, standalone sellers like TwitterAPI.io at $0.15 per 1K tweets or Apify actors from $0.00025 a tweet. For a single public page inside an AI agent, FATSEARCH reads the page through your own provider key without a scraping API. There is no best for all cases.

### Is there a free Twitter API?

Not from X: its API is pay-per-use, billed per post read. Some RapidAPI listings have a free plan with a small request cap, and Apify has a free plan too, so you can test before paying. Open-source tools (twscrape, twikit) are free code, but need your own X accounts and break when X changes its internal API.

## Related

- [Web scraping MCP: every reader we measured](https://fatsearch.com/compare/)
- [Benchmarks: every run, charted](https://fatsearch.com/benchmarks/)
- [Method: one page, six readers](https://fatsearch.com/docs/benchmarks/#one-page-six-readers)
- [Method: the hard-pages run](https://fatsearch.com/docs/benchmarks/#other-readers)

By FATSEARCH · data as of 2026-10-19 · reviewed 2026-10-19 · https://fatsearch.com/guides/how-to-scrape-twitter/