The 9 Best Pocket Alternatives in 2027 (And What To Do If You Never Exported)

The best Pocket alternative in 2027 is Marqly for most people: it imports the list.csv from your Pocket export archive and then adds the one capability Pocket never shipped — finding a save by describing it rather than remembering its title. Raindrop.io is the best free option, with unlimited bookmarks and collections on every platform and no expiry. Wallabag is the best pick if the reason you are here is that a vendor disappeared, because it is open source under MIT, self-hostable, and has been maintained continuously since 2013.

The conversation around this query has changed shape. In 2025 and through most of 2026 the question was how do I migrate off Pocket while it still exists. In 2027 Pocket has been gone for a year and a half, its export window closed on November 12, 2025, and the people searching are split into two very different groups: those holding an old export archive who need somewhere to put it, and those who never exported and are finding out whether it is recoverable. This guide answers both, and it now scores tools on a criterion neither group was asking about two years ago — whether the tool will still exist in 2029.

That criterion is not paranoia. Two of the most-recommended read-it-later tools in the world shut down within eight months of each other, and one of them was free.

How this list was compiled. We compare import formats, capture, organisation, retrieval, pricing and durability using dated competitor fact sheets and linked official sources, and we publish the criteria in how we test. Marqly publishes this article and is one of the products discussed, so weigh our own section accordingly. Our separately measured import results live in the import-fidelity study, which is real test data rather than a claim. Where a competitor wins a row, we say so.

What happened to Pocket, exactly

Mozilla announced Pocket’s shutdown in May 2025 and turned off the service on July 8, 2025. The timeline that matters in 2027:

1. July 8, 2025 — shutdown. Mozilla ended the Pocket service and turned off new saves. Apps and extensions stopped syncing.
2. November 12, 2025 — export window closed. Export access and the API were disabled, and Mozilla says remaining data was queued for permanent deletion from that date.
3. Premium subscribers received prorated refunds for remaining subscription time.
4. 2027 — the archive-rescue phase. If you exported, your library is a ZIP on a disk somewhere and the tools below will read the CSV inside it. If you did not, there is no supported recovery path.

Source: Mozilla’s Pocket shutdown notice, verified October 6, 2026.

Worth remembering what you are replacing, because it explains the shape of this market. Pocket’s search was title-and-URL only unless you paid for Premium full-text search. Tagging was fully manual, with suggested tags a Premium extra. Highlights were capped at three per article on the free plan, with no notes or annotations. Pocket was pleasant to save into and close to useless to retrieve from — which is why “I have 4,000 saves and cannot find any of them” was the single most common complaint, and why semantic search is now the main axis this list is sorted on.

What to do in 2027 if you never exported

This is the part most Pocket-alternative articles skip, and in 2027 it is half the audience. Check these four places in order before you accept the loss:

1. Your downloads folder and cloud drives. Look for a ZIP with ril_export in the name, or a folder containing list.csv. Pocket’s export produced an archive, and most people downloaded it to a default location and forgot. Search your file system for list.csv and for .zip files created around the middle of 2025.
2. Any tool you already synced Pocket to. Readwise, IFTTT-style automations, Notion databases and some RSS bridges pulled Pocket saves on a schedule. If you ever connected Pocket to anything, that thing may hold a partial copy.
3. Your browser’s own bookmark file. If you ever ran a Pocket-to-browser export, Chrome, Edge, Firefox and Safari each hold a standard HTML bookmark file you can export today — and that file imports into every tool below.
4. Your email. Search for Pocket or Mozilla messages from mid-2025. Shutdown notices sometimes contained a direct export link, and while those links are dead, the email may confirm whether you completed a download and when.

If all four come up empty: Mozilla states that export access was disabled on November 12, 2025 and that data deletion began from that date, so there is no supported recovery. The practical move is to start fresh with a tool whose export path you have tested on day one — see the durability column below — and to treat the loss as the argument for that habit rather than as bad luck.

Two shutdowns in eight months: how to pick a tool that survives

Omnivore shut down in November 2024 after its team was acqui-hired by ElevenLabs. Users had until November 30, 2024 to export before deletion. Omnivore was completely free, with no paid plan at any point — and that free-with-no-revenue model is precisely what made the shutdown possible. The open-source code remains on GitHub and community forks offer self-hosted deployments, but the forks are volunteer-maintained with no guarantees, and self-hosting the original stack is genuinely hard (PostgreSQL, Elasticsearch and multiple microservices).

Pocket shut down on July 8, 2025. Eight months later. Different owner, different reason — a portfolio cut rather than a failed business — and the same outcome for anyone who had not exported.

The lesson for choosing in 2027 is not “avoid free tools.” It is that three properties determine whether your library outlives the vendor, and you can check all three in ten minutes:

• Can you get everything out, in a documented format, without asking anyone? Wallabag, Obsidian and the self-hosted pair are strongest here by construction. Marqly exports CSV carrying URL, title, description and tags, with a complete account copy including highlights, notes and boards available from support. Raindrop exports cleanly.
• Is there a revenue model that does not depend on you being the product or on a single acquirer’s mood? Paid plans with real subscribers (Raindrop, Readwise, Marqly) and long-lived open-source projects with a hosted option funding development (Wallabag since 2013) score better than free-with-no-revenue.
• Is the code open source, so a shutdown leaves a fork rather than a hole? Wallabag (MIT), Linkwarden (AGPL-3.0) and Karakeep all survive their company. Proprietary tools do not.

How the best Pocket alternatives compare

Tool	Best for	Free tier	Semantic search	Imports the Pocket CSV	Export path	Price (verified Oct 2026)
Marqly	Retrieval by meaning	100 bookmarks	✅ Pro	✅ list.csv	CSV; full copy via support	Free; Pro $72/yr or $9/mo
Raindrop.io	Free general-purpose saving	✅ Unlimited	❌ Keyword	✅ Yes	Collections export	Free; Pro $28/yr
Wallabag	Durability + clean reading	✅ Self-hosted	❌ None	✅ Yes	Open formats, MIT	Free; €11/yr hosted
Karakeep	Self-hosted AI tagging	✅ Self-hosted	✅ BYO key	⚠️ Via standard import	Your own database	Free; Cloud Pro $4/mo
Linkwarden	Full-page archiving	✅ Self-hosted	❌ Optional AI tags	✅ Yes	Your own database	Free; Cloud $3/user/mo
Instapaper	Minimalist reading	✅ Unlimited saving	❌ None	✅ Yes	CSV/HTML export	Free; Premium $5.99/mo or $59.99/yr
Readwise Reader	Highlights + power reading	❌ No permanent tier	❌ Keyword	✅ Yes	Highlight export	$9.99/mo annual ($119.88/yr)
Matter	Reading + listening	⚠️ Limited	❌ None	✅ Yes	Limited; no public API	$8/mo or $60/yr on web
Notion Web Clipper	Already living in Notion	✅ Clipper free	⚠️ Business plan only	⚠️ Via Notion import	Markdown/CSV export	Free; Plus $10, Business $20/member/mo

A note on the import column: “imports the Pocket CSV” means the tool accepts the list.csv from a Pocket export archive and re-saves those links into your new library, carrying titles, URLs and tags. Original save dates are generally not preserved — importers read the date column but discard it on insert — and no importer brings back highlights or article text, because a Pocket export never contained them.

One notable exclusion: mymind does not belong on this list despite being an otherwise good visual saver, because it has no bulk import by design. You cannot bring a Pocket archive into it, and for a Pocket refugee that is the whole job. Its free guest plan also caps at 100 cards.

1. Marqly — best for making a dead library findable again

Pocket’s core failure was that saving was easy and finding was hard. Marqly is built as the direct answer: semantic search retrieves a save by meaning, so “that article about sleep and cortisol” works even when you have forgotten the title and the exact words. On Pro it also auto-tags every save, generates AI summaries, and answers questions across your whole library through Ask, with numbered citations back to the source.

For a Pocket refugee specifically, the import path matters more than the feature list. Marqly reads the list.csv inside the Pocket export archive — not the HTML preview file, whose <ul><li> layout is not the Netscape bookmark format and does not parse (we measured this: 0 of 261 items from a genuine export; see the import-fidelity study). Titles, URLs, descriptions and tags come across; folder structure flattens to at most two board levels; original save dates do not. Auto-tagging the import is Pro, so on the free plan you land with your existing tags intact and no new ones.

The honest limits are the ones a Pocket user will feel first. The free tier stores up to 100 bookmarks, so a 4,000-item archive needs Pro from day one. There is no AI on free at all — auto-tagging, summaries and semantic search are every one of them Pro. There is no public API and no self-hosting, so the durability argument above is answered by CSV export and a support-provided full account copy rather than by owning the code. Offline reading is Pro, and it is per-device on the web and iOS apps rather than a synced offline library.

Who it’s best for: anyone whose Pocket library became a graveyard — hundreds or thousands of saves, none findable — who wants the pile searchable this afternoon. Skip it if you want a bare text reader with no AI (Instapaper is cheaper), you want to own the code (Wallabag, Linkwarden, Karakeep), or you need a free tier that holds an unlimited library (Raindrop.io).

• Pros: semantic search across titles, content, highlights and transcripts; auto-tagging and AI summaries on Pro; imports the Pocket list.csv, Raindrop HTML and browser bookmark HTML; six-colour highlighter free on every plan; YouTube AI card keeps transcripts with the video; available on web, Chrome, Edge, Firefox, Safari, iOS and Android.
• Cons: free tier capped at 100 bookmarks; no AI of any kind on free; no public API or self-hosting; offline is Pro and per-device; auto-tagging an import requires Pro.
• Price: free tier, no card required · Pro $72/year (about $6/month billed annually) or $9/month · standing $49 first-year offer · verified students $48 first year.

Get started free →

2. Raindrop.io — best free, general-purpose replacement

Raindrop is the most common straight-across Pocket replacement and the easiest to recommend on price alone: unlimited bookmarks, unlimited collections, tags and every app — web, Mac, Windows, Linux, iOS, Android, and extensions for Chrome, Edge, Firefox and Safari — free, with no expiry and up to three highlights per bookmark. Public collections share a curated list from a link. It saves articles, videos, PDFs, images and whole pages, and it is a mature, stable product.

Where it stops short of the 2027 frontier is retrieval. Search on free is titles and tags; full-text search requires Pro, and even then it is keyword-shaped rather than semantic. If your Pocket complaint was “I can never find anything I saved,” Raindrop will not fix that — it will give you a much nicer place to lose things. Pro at $28/year or $3.25/month adds full-text search, Stella AI save-time tag and collection suggestions, permanent page copies, annotations and daily backups, and permanent page copies are the strongest link-rot defence among the hosted options.

Who it’s best for: people who want a free, dependable Pocket-shaped tool and do not care about AI. Skip it if retrieval is your actual pain, because no price tier changes the keyword-only model.

• Pros: the most generous free tier here; apps on every platform including Linux; polished visual organisation with nested collections; Pro is cheap and adds permanent page copies.
• Cons: no semantic search at any price; full-text search is paywalled; organisation is entirely manual.
• Price: free (unlimited) · Pro $28/year or $3.25/month.

3. Wallabag — best if the reason you are here is that a vendor vanished

Wallabag is open source under MIT, free to self-host, and has been maintained continuously since it started in France in 2013. That combination is the direct answer to the Pocket and Omnivore lesson: there is no company that can decide to shut it down, and if the project ever stalls, a working copy is already on your server. It is a read-it-later tool first, which means it reproduces the behaviour Pocket refugees actually miss — save a page, read a clean distraction-free copy later.

If you do not want to run a server, official wallabag.it hosting is €11/year with all features included, plus €4 for 3 months and a €30/year supporter plan that funds development. That is the cheapest paid option in this article and, unlike a free tier, it has no cap.

The trade is that Wallabag has no AI of any kind: no auto-tagging, no summaries, no semantic search. Organisation and retrieval are on you, and it is weaker as a general bookmark organiser than as a reading queue.

Who it’s best for: anyone whose primary 2027 criterion is “this will still exist,” plus technical users happy to self-host. Skip it if you want AI retrieval or a polished native app experience.

• Pros: MIT open source, free to self-host, 13-year continuous maintenance history; clean read-it-later rendering; hosted option at €11/year with everything included; no proprietary format.
• Cons: no AI features at all; self-hosting is real setup and maintenance; smaller app ecosystem; reading-focused rather than organiser-focused.
• Price: free self-hosted · €11/year hosted · €4 for 3 months · €30/year supporter.

4. Karakeep — best self-hosted option that actually has AI

Karakeep (formerly Hoarder, renamed in 2025) is the answer if you want both properties at once: your own server and automatic organisation. Self-hosting is fully free and open source with all features included, and an LLM you configure — an OpenAI-compatible key or a local Ollama model — auto-tags everything, with full-text and semantic search across it all. Extensions for Chrome, Firefox and Safari plus iOS and Android apps make it unusually complete for a self-hosted project, and local-LLM support means the tagging can be private as well as free.

For a Pocket rescue, the caveat is that Karakeep’s import path is a standard bookmark import rather than a Pocket-specific one, so plan on converting your list.csv to a standard format first. The hosted Karakeep Cloud is in public beta with a $0 tier capped at 10 bookmarks and a $4/month Pro tier — ten bookmarks is a preview, so “free Karakeep” means your own server.

Who it’s best for: technical users who want AI organisation without handing a vendor their library. Skip it if you do not want to administer software, or if AI costs (API fees or local GPU) are a problem.

• Pros: AI tagging and semantic search, free, under your own roof; local-LLM support for private AI; full apps and extensions; a cheap $4/month cloud Pro as an exit from self-hosting.
• Cons: you run the server, updates and backups; AI costs surface eventually; the cloud free tier caps at 10 bookmarks; Pocket import needs a conversion step.
• Price: free self-hosted · Cloud Pro $4/month (annual price not published).

5. Linkwarden — best for archiving so link rot cannot win

Linkwarden is open source under AGPL-3.0 with all features included when self-hosted, and it is built around full-page preservation: every bookmark gets an archived copy. For a library that has already survived one vendor’s disappearance, that is a different kind of insurance — it protects against the pages dying, not just the service. A 2019 Pocket archive is full of dead URLs by 2027, and Linkwarden is the only tool here that would have kept the content.

It is collaborative, supports highlights and RSS, has real mobile apps and browser extensions, and offers optional AI tagging if you wire up a provider. The cloud plan is $3/month per user billed yearly (25% off) or $4/month billed monthly, supports up to 30,000 links per user, and has no free tier beyond a trial — so free Linkwarden means self-hosted Linkwarden.

Who it’s best for: preservationists, researchers, and anyone whose old archive is already full of dead links. Skip it if AI retrieval is the priority (Karakeep is stronger there) or you do not want to run a server.

• Pros: free forever self-hosted with every feature; full-page archives of everything — the best link-rot answer here; active development with mobile apps and extensions; collaboration built in.
• Cons: “free” costs a server and your maintenance time; AI is an optional bolt-on; cloud has no free tier beyond a trial; setup is real work.
• Price: free self-hosted · Cloud $3/user/month billed yearly or $4/month billed monthly.

6. Instapaper — best minimalist reader, but the price doubled

Instapaper is the elder statesman of read-it-later and it is still here, which in 2027 is itself a data point. If all you want is to save articles and read them later in a clean, fast, text-only view, Instapaper nails it, and its free tier covers unlimited saving and offline reading — genuinely usable rather than a demo.

Two corrections matter here, because a lot of older advice is stale. Premium is $5.99/month or $59.99/year — it roughly doubled in late 2023 from $3/month and $30/year, without major new features at the time, so any article still quoting “$3/month” is describing a price that no longer exists. And free accounts cannot full-text search their own saves; search is paywalled, notes are capped at five per month on free, and there is no AI at all — no auto-tagging, no summaries, no semantic search, with organisation by manual folders. Premium adds full-text search, a permanent archive, unlimited notes, text-to-speech playlists, speed reading and send-to-Kindle.

Who it’s best for: minimalists who read within a week of saving and want the least app possible. Skip it if you are rescuing a large old archive, because there is no AI to help you dig through it — see our dedicated Instapaper alternatives roundup if you are leaving Instapaper rather than arriving at it.

• Pros: clean fast reader; free tier includes unlimited saving and offline reading; reliable and long-lived; simple enough to never need training.
• Cons: full-text search is paywalled; free caps notes at five per month; no AI of any kind; Premium roughly doubled in price in late 2023; development has been quiet for years.
• Price: free · Premium $5.99/month or $59.99/year.

7. Readwise Reader — best for power readers who highlight

Reader is the premium choice for serious readers: articles, PDFs, email newsletters and ebooks, with best-in-class highlighting and a spaced-repetition review system that resurfaces your highlights over time — a genuinely different idea from anything Pocket offered. AI features cover summaries and YouTube handling, and highlights flow into Notion or Obsidian.

The catch is price and the absence of a free option. There is no permanent free tier — after a 30-day trial you must subscribe at $9.99/month billed annually ($119.88/year) or $12.99/month monthly for the Full plan. Reader is only in Full, so the cheaper Lite plan at $5.59/month billed annually covers classic Readwise highlight syncing without the Reader app at all. Search is keyword-based rather than semantic, tagging is manual, and there is no public sharing of collections.

Who it’s best for: researchers, students and writers who highlight obsessively and want those highlights resurfaced. Skip it if you want to save and find articles cheaply, or if you need free. See Readwise Reader vs Marqly for the head-to-head on retrieval.

• Pros: the most complete reading pipeline; outstanding highlighting plus spaced-repetition review; handles every format; highlights export into Notion or Obsidian.
• Cons: no permanent free tier; the most expensive option here; Reader requires the Full plan; keyword rather than semantic search; manual tagging; steeper learning curve.
• Price: no free tier · Full $9.99/month billed annually or $12.99/month monthly · Lite $5.59/month billed annually.

8. Matter — best for reading and listening, with an App Store pricing trap

Matter pairs a pleasant reader with high-quality text-to-speech, so you can listen to saved articles like a podcast. If audio is how you consume long reads, the TTS is genuinely good and nothing else on this list competes with it. Premium unlocks HD voices, unlimited highlighting and notes, full-text search and integrations.

Two things to check before subscribing. Pricing differs by channel: $8/month or $60/year on the web, versus $7.99/month or $79.99/year through the iOS App Store — subscribing in the app costs $20 more per year for the same plan, which is an easy trap. A higher Matter Pro tier at $12.99/month also appears in the App Store. And there is no Android app; Matter is built for iPhone, iPad, Mac and the web, with no public API and no public sharing of collections.

Who it’s best for: commuters and multitaskers whose queue is mostly long reads they would rather hear. Skip it if you need deep search, a big organised archive, or Android.

• Pros: excellent text-to-speech; pleasant reader; good for audio-first consumption; web pricing is reasonable.
• Cons: no Android app; App Store annual pricing is $79.99 versus $60 on the web; no public API and no public sharing; unlimited highlighting, HD voices and full-text search all sit behind Premium.
• Price: $8/month or $60/year on web · $7.99/month or $79.99/year via iOS App Store · Matter Pro $12.99/month in the App Store.

9. Notion Web Clipper — best if you already live in Notion

If your notes, projects and tasks already run in Notion, the Web Clipper keeps everything in one place: it is free, works with any Notion account including the generous free plan, and clips land in databases, so saved links inherit Notion’s properties, views, filters and tags. Offline mode arrived in 2025 on all plans, and there are native apps for web, Mac, Windows, iOS and Android.

For a Pocket rescue the caveats are real. The clipper is bare-bones at capture time — no tags, notes, highlights or screenshots, just a destination picker — and Notion’s own documentation warns that clip fidelity varies and formatting can go missing. Importing a Pocket archive goes through Notion’s general import rather than a Pocket-specific path. And the AI everyone assumes is there is a Business-tier capability: meaningful AI search and chat over clipped content is bundled with Business and above at $20/member/month, while Free and Plus get only a trial of it. Official browser support has narrowed to Chrome and Safari.

Who it’s best for: Notion-native users who want resources beside their notes and are already paying for Business. Skip it if you want a dedicated reader, reliable capture of formatted pages, or AI without the top tier — our Notion Web Clipper alternatives guide goes deeper, as does Marqly vs Notion Web Clipper.

• Pros: clipper is free and works on the free plan; database properties give views and filters; one workspace for notes, tasks and saves; offline on all plans; official API and mobile share-sheet saving.
• Cons: no tags, notes or highlights at capture time; documented clip-fidelity problems; AI search needs the $20/member/month Business plan; browser support narrowed to Chrome and Safari; Pocket import is generic rather than purpose-built.
• Price: clipper free · Plus $10/member/month · Business $20/member/month (billed annually).

Which Pocket alternative should you choose?

• You want the pile to be findable again → Marqly. It imports the list.csv and adds the semantic search Pocket never had.
• You want free and unlimited → Raindrop.io.
• You want the tool to outlive the next shutdown → Wallabag (MIT, since 2013) or Linkwarden (AGPL-3.0, full-page archives).
• You want AI and full data ownership → Karakeep, self-hosted.
• You want the least app possible → Instapaper, and accept that search costs Premium.
• You highlight everything → Readwise Reader, and accept the price.
• You would rather listen → Matter, and subscribe on the web.
• You already live in Notion → Notion Web Clipper, and check whether you have Business.

Match the tool to your actual pain, not its feature list. The most common mistake in a post-shutdown migration is optimising for capture — a nicer save button — when the reason 4,000 Pocket saves felt useless was retrieval.

How to actually move your Pocket archive

1. Find the archive. Locate the ZIP Pocket produced. Inside you should see a CSV (commonly list.csv) and an HTML preview file. If you only have the HTML file, see step 2.
2. Use the CSV, not the HTML. The HTML preview is a <ul><li> layout, not the Netscape bookmark format, and importers that expect Netscape will not parse it — we measured 0 of 261 items from a genuine export. If the CSV is missing, convert what you have with the free Pocket Export Converter and inspect the result in the Bookmark File Viewer.
3. Clean before you import. A multi-year Pocket library is full of duplicates and dead links. Run it through the free Duplicate Bookmark Finder and Dead Link Checker first; importing 4,000 links of which 900 are dead just relocates the problem.
4. Import, then verify. Marqly’s importer takes .html/.htm/.csv files up to 10 MB on free and 30 MB on Pro, with 10,000 bookmarks per file. Titles, URLs, descriptions and tags carry across; folder structure flattens to at most two board levels; original save dates do not — imported items take the import date.
5. Tag once, or let Pro do it. Auto-tagging an import is a Pro feature. On free, your existing Pocket tags come across and nothing new is added, so budget an evening if you want the archive organised by hand.
6. Test the exit on day one. Export something back out immediately and confirm the file is readable. That single habit is the actual lesson of Pocket and Omnivore.

Follow the Pocket migration guide for the tool-specific steps, or start from the universal Migration Center. All of the cleaning tools above are in the free tools directory.

What actually matters in 2027

Retrieval beats capture. Pocket proved that a delightful save button plus title-only search produces a library nobody uses. If you take one thing from this list, take that the differentiating feature in 2027 is finding things, and the only capability that meaningfully changes finding things is semantic search — Pro in Marqly, absent from Raindrop and Instapaper at any price, and available self-hosted in Karakeep.

Durability is now a feature you can evaluate. Two major tools died in eight months. Check the export path, the revenue model and whether the code is open source before you commit four thousand links to anyone. Wallabag, Linkwarden and Karakeep are strongest on that axis by construction; every hosted tool in this table is weaker, including ours.

Link rot is the slow failure. An archive rescued in 2027 is full of pages that died between 2019 and now. Full-page archives (Linkwarden), permanent copies (Raindrop Pro, Instapaper Premium) and local save-as-PDF (Marqly, processed in Chrome and Edge) are the three real defences; most tools offer none.

If the reason you are reading this is a ZIP file you have not opened in a year, Marqly is built to make that pile findable — import the CSV, let Pro tag and summarise it, and start searching everything you have ever saved by meaning rather than by title. Compare it directly in Pocket vs Marqly, read the measured import-fidelity study, or see the free-tier roundup if budget is the constraint.

---

Prices, plan limits and shutdown dates verified against official sources in October 2026; the Pocket timeline was checked against Mozilla’s notice on October 6, 2026. Changes from the 2026 edition: Instapaper Premium corrected from a stale “~$3/month” to $5.99/month or $59.99/year (it doubled in late 2023); Marqly’s platform list corrected to include Edge, Firefox, Safari and Android; mymind removed from the ranked list because it has no bulk import by design; Karakeep and Linkwarden added; Matter’s App Store pricing trap documented; Notion’s AI clarified as a Business-tier capability; and a durability criterion added after the Omnivore and Pocket shutdowns. We’ll update this guide as plans change.