ArchiveBox Alternatives for Self-Hosted Web Archiving

ArchiveBox is a free, open source program that saves copies of the URLs you feed it on your own machine. It is actively developed in November 2026. Version 0.9.36, released September 22, brought a new archiving engine, PostgreSQL support, scheduled crawls and more than 50 new plugins, and point releases have followed since, up to 0.9.77 on October 27, 2026. The upgrade moves your data folder to a new layout, and the release notes tell you to make a full backup first.

People still look elsewhere for practical reasons. ArchiveBox saves each URL several ways, from SingleFile HTML and PDF to a screenshot, WARC and yt-dlp media, so every snapshot means a browser run and a stack of extractors. An upgrade bug that could pin the background runner at 100% CPU was only fixed in v0.9.70. By default it also submits each URL to archive.org, which you have to turn off for a private list. Others want phone apps and a nicer place to read what they saved.

The picks fall into four groups. Linkwarden and Karakeep are bookmark managers that archive. SingleFile and ArchiveWeb.page save pages from your browser. Browsertrix and Wget crawl. Readeck and wallabag are read-it-later apps that store the text. For copying whole sites, see our HTTrack alternatives. For old copies that someone else made, see Wayback Machine alternatives.

Prices below are as of November 6, 2026. Our method is on How we review tools.

Quick comparison

Linkwarden

Of everything here, Linkwarden comes closest to the ArchiveBox habit of saving each page more than one way. Every link gets a screenshot, a PDF and a single HTML file, and there is a reader view for highlighting and notes. Sending a copy to the Wayback Machine is an option, not a requirement.

The front end is where it pulls ahead. You can share collections with other people, save from native iOS and Android apps, a browser extension or an iOS Shortcut, and search across what you kept. The AGPL-3.0 repo has about 20.2k stars. Release v2.16.3 on September 9 sped up link, search, tag and collection queries on large libraries.

Best for: Self-hosters who want ArchiveBox-style copies inside a bookmark manager that a family or team can share.

Pricing:

• Self-hosted: free, open source under AGPLv3.
• Cloud: $3 per user per month, billed yearly, which the page lists as 25% off monthly billing. Up to 30,000 links per user, 14-day free trial.

Watch out for:

• Preservation is the heavy part. The docs say a fresh instance idles at about 350 MB since v2.15, that memory settles higher when it archives regularly, and that Monolith and PDF are the most expensive formats.
• Turning browser tasks off to save memory removes screenshots, PDFs, HTML snapshots and the reader view in one go, which leaves a plain bookmark list.

Official site: Linkwarden.

Karakeep

Formerly Hoarder, Karakeep is a bookmark app built around finding things again. It fetches titles and images, tags and summarizes links with an LLM (OpenAI or a local model through Ollama), reads text out of images with OCR, and offers full-text and semantic search. It can import from Chrome, Pocket, Linkwarden and Omnivore, and pull new items from RSS feeds.

Archiving is there but switched off by default. Full local copies through monolith and video downloads through yt-dlp both need a setting changed, and out of the box it keeps readable text and a screenshot. Release 0.33.2 from August moved to a Chrome 151 container and added CLI upload of SingleFile archives. The AGPL-3.0 project has about 29.9k GitHub stars and apps for iOS and Android.

Best for: People with thousands of links who care more about search and automatic tags than about perfect page copies.

Pricing:

• Self-hosted: free under AGPL-3.0.
• Cloud Free: $0, 50 bookmarks, 100MB.
• Cloud Pro: $4/month or $40/year, 50,000 bookmarks, 50GB, AI tagging, 7-day money-back guarantee.

Watch out for:

• The Docker guide calls an OpenAI key "highly recommended" for automatic tagging. The other route is running Ollama on your own hardware.
• If you came for archives, set the full-page archive option before you import. It defaults to false, so earlier bookmarks have text only.

Official site: Karakeep.

SingleFile

Skip the server entirely and SingleFile still covers the most common archiving job. Click the extension and the page you are reading becomes one HTML file, images, fonts and styles included, that opens in any browser. ArchiveBox uses the same tool for one of its outputs, so the files look familiar.

The Firefox add-on is at 1.35.0, updated November 4, 2026, with about 93,000 users and a 4.8 rating from 1,054 reviews. Because it saves what your browser already rendered, pages behind a login come out right without handing cookies to a server. The CLI version drives Chrome or Chromium for batch jobs and crawls.

Best for: Saving articles, receipts and docs pages one at a time, with no software to host.

Pricing:

• Free and open source under AGPL-3.0. Commercial licences are available on request.

Watch out for:

• Canvas and video elements sometimes cannot be saved because of browser security limits.
• There is no library. Files land in your downloads folder, so search and tagging are up to you or a tool like Karakeep.

Official site: SingleFile.

ArchiveWeb.page

ArchiveWeb.page records rather than fetches. You start a session, browse normally, and every page you open goes into a private archive that you can view offline and export as WARC or WACZ. ReplayWeb.page plays those files back with menus, scrolling and clicks intact.

WARC is also what ArchiveBox writes, so captures can move between tools and into institutional archives later. Autopilot scrolls infinite feeds and runs scripted behaviors for Twitter, Instagram and Facebook. It ships as a Chromium extension or a desktop app for Windows, macOS 10.15+ and Linux, and v0.16.2 came out on April 30, 2026.

Best for: Researchers and journalists who need a faithful, replayable record of pages they had to click through.

Pricing:

• Free and open source under AGPL-3.0.

Watch out for:

• The extension needs Chrome or another Chromium browser, and for whole sites the project points you to Browsertrix.
• The docs warn that Autopilot behaviors "are not always guaranteed to work as expected".

Official site: ArchiveWeb.page.

Browsertrix

Webrecorder built Browsertrix for the crawling that ArchiveWeb.page leaves out. It loads every page in a real browser, so script-heavy sites that a plain fetcher misses still come through. You set the crawl in a web dashboard instead of a config file.

The hosted service charges by execution minutes. Starter includes 180 minutes, 100 GB of storage and 2,000 pages per crawl, and only that plan has a free trial. The code is AGPLv3 for anyone who wants to run it in-house.

Best for: Archivists and teams who need high-fidelity copies of large or dynamic sites and would rather not run a crawler.

Pricing:

• Starter: $30/month, 180 execution minutes, 100 GB, 2,000 pages per crawl, 1 concurrent crawl, 7-day free trial.
• Standard: $60/month, 360 minutes, 220 GB, 5,000 pages per crawl, 2 concurrent crawls.
• Plus: $120/month, 720 minutes, 500 GB, 10,000 pages per crawl, 3 concurrent crawls.
• Pro (from 2 TB) and Enterprise: billed annually, no public price.

Watch out for:

• Running your own copy means Kubernetes, which is a big step up from the single Docker Compose file most ArchiveBox users know.
• Minutes run out. Extra execution time on Starter is sold on top of the plan.

Official site: Browsertrix.

GNU Wget and Wget2

ArchiveBox calls wget for one of its own outputs, so you can go to the source if a terminal and a cron job are all you need. One command with --page-requisites and --convert-links saves a page plus the images and CSS it needs, and --warc-file writes a WARC next to it.

Classic Wget is at 1.25.0 from November 10, 2024. Wget2 reached 2.3.0 on September 21, 2026 and adds threads, HTTP/2 and parallel connections. Both are in the package managers for Linux, macOS, Windows and BSD.

Best for: Admins who want a scripted, dependency-free capture of a list of URLs, with WARC output and no web interface.

Pricing:

• Free software under the GNU GPL.

Watch out for:

• It reads links from HTML and CSS only, so content that JavaScript builds in the browser is missing from the copy.
• Wget2 drops FTP and some older options, so check old scripts before you move them over.

Official site: GNU Wget and Wget2.

Readeck

Readeck is a read-it-later app first and an archive second. The text and images of a page are stored on your instance the moment you save a link, and each bookmark is kept as a single, immutable ZIP file. You get labels, highlights, saved searches as collections, video transcripts and ebook export of whole collections.

It is the lightest server on this list. Readeck ships as one executable with no dependencies for Linux, FreeBSD, macOS and Windows, or as a Docker or Podman image. The docs ask for 512 MB of memory and about 200 MB of disk plus roughly 300 KB per bookmark. Version 0.23.4 is dated September 17, 2026.

Best for: Readers who mostly save articles to read later and want a tiny server that keeps the text if the original disappears.

Pricing:

• Free and open source under AGPL-3.0. The project page describes self-hosting only and lists no paid plan.

Watch out for:

• Pages the server cannot reach, such as ones behind your logins, need the Firefox or Chrome extension, which sends what your browser can see.
• Bookmarks are stored as ZIP files of the readable content, not as WARC, so they are not a pixel-perfect record of how the page looked.

Official site: Readeck.

wallabag

Open source since 2013, wallabag is the veteran of self-hosted read-it-later. It strips saved articles down to text you can read on the web, in its mobile apps or on an e-reader, and the latest release is 2.6.14. ArchiveBox can import a wallabag list, so the two can run side by side.

If you are tired of running servers, wallabag.it hosts it for you. A year costs 11 euros, plans do not renew on their own, and the 14-day trial needs no card.

Best for: People whose archive is really a reading list, and who want a hosted option for less than a euro a month.

Pricing:

• Self-hosted: free and open source.
• wallabag.it: 4 euros for 3 months, 11 euros for 1 year, or a 30 euro yearly supporter plan. Prices are in euros only. Email saving, RSS feeds and imports are for paying subscribers.

Watch out for:

• When the server fetches the URL, paywalled and JavaScript-rendered pages fail. Maintainers say only the desktop extension sends the rendered HTML, with no equivalent on Android or iOS.
• It keeps the extracted article, not screenshots, PDFs or WARC files. When extraction grabs the wrong parts, the fix is a hand-written site config with XPath rules.

Official site: wallabag.

How to choose

When to stay with ArchiveBox

Keep ArchiveBox if you want many formats of every URL, WARC included, with schedules and imports from browser history, bookmarks and RSS. None of the bookmark apps above write WARC, and Wget gives you folders, not a library to browse. The 0.9 line also added PostgreSQL, scheduled crawls and a REST API, so many of the old pain points are being fixed. Back up before the upgrade and turn off archive.org submission if your list is private.

Frequently asked questions