An honest comparison of X bookmark managers
We build MarkVault. Weigh everything below knowing that. To keep it useful anyway: every claim about another product comes from that product's own website, pricing page or FAQ, checked on 2026-08-20 and listed in Sources at the bottom. There is a section on what each of them does better than us, and a section on where we fall short. If something here is wrong or out of date, tell us and we will fix it — this page is only worth anything if it stays true.
The short answer
If you want the most mature product with the biggest user base, that is Twillot. If you want it on your iPhone today, that is XSaved. If you save from LinkedIn and TikTok as well as X, that is Dewey. If you want your bookmarks turned into a reading digest, that is Tweetsmash.
MarkVault is for one specific person: someone with years of X bookmarks, a lot of them in Chinese, who wants them as files on their own disk rather than in someone's product, and who would rather pay once than subscribe. If that is not you, one of the others is probably a better fit, and we would rather you find that out here than after paying us.
Side by side
Competitor figures verified 2026-08-20 from their own public pages. XSaved made its Chrome extension free and its pricing page was rechecked 2026-09-09; the row on where its data lands comes from unpacking its store build that same day, not from its marketing copy. Prices change; if you are reading this much later, check the source links.
| MarkVault | Twillot | XSaved | Dewey | Tweetsmash | ContextBolt | |
|---|---|---|---|---|---|---|
| Billing | One-time $29 / $39 | Subscription, 4 tiers by stored count | Free for Chrome — Pro is sold in the iPhone app | $10/mo · Lifetime $225 | $7/mo · Lifetime $198 | $6/mo |
| Free tier cap | Unlimited | 1,000 posts | Unlimited (local) | Manual sync only | — | 150 bookmarks |
| Export | Free — it is the storage format | Free | Free, forever | $50 pass, valid 48h, or yearly plan | $49 one-time pass | Pro only |
| Where data lands | Plain text files in a folder you pick + IndexedDB | “In your own browser” — form unspecified | IndexedDB in the browser — sign in and your post text is mirrored to their backend | Not stated | On their servers | Browser, but post text is sent to them for tagging |
| Account required | No | No | No (sync needs one) | Yes | Yes | Yes |
| Where search runs | Your browser — tested on a real 38,000-post archive and a synthetic 377,000 | Their site | Your browser | Their servers | Their servers | Your browser |
| Search that handles Chinese | Yes, ranked by relevance | No — its own in-app help says a space means or and it matches literal text | Barely — its index drops Chinese characters, so a plain query is not filtered by them; searching inside a folder matches the raw text, and a quoted query does so only for very short phrases | Search runs on their servers — we cannot check it | Search runs on their servers — we cannot check it | Yes, but only as raw text — no word splitting; separate words are matched one by one and ordered by how many appear, then by date |
| Chinese interface | Yes | Yes | No | No | No | No |
| Shipping today | Not yet — late Sept 2026 | Yes, since 2023 | Yes | Yes | Yes | Yes |
What each of them does better than us
Twillot
Three years of shipping, 10,000+ installs, and a Chinese-language community we have nothing comparable to. Auto-filling folders, media downloads, and a suite of free standalone tools. If you want a product with a track record, this is it.
XSaved
iOS and Mac in beta — we have no mobile story at all. Notes attached to bookmarks, passkey sign-in, opt-in sync across devices. Their own comparison page is the most honest piece of writing in this category and it set the standard for this one. And if your archive is mostly English, their free whole-word search is usually enough — our free tier gives you none.
Dewey
Breadth we do not attempt: LinkedIn, Bluesky, TikTok, Threads and web bookmarks in one place, Notion sync, RSS output, public shareable collections.
Tweetsmash
A completely different job — turning saves into scheduled reading digests, with Notion and Zotero sync and an official-API import path. If you want to consume your bookmarks rather than archive them, we do not compete.
ContextBolt
An MCP endpoint that works today with Claude, Cursor and Codex, plus AI auto-tagging. We have neither yet.
Where MarkVault falls short — read this part too
- We have not launched. Target is late September 2026. Every product above is shipping today with real users. What you are reading is a promise, not a record.
- No users, no reviews, no track record. Buying at the founding price means betting on an unproven product from an unproven developer.
- Chrome only. No Firefox, no Safari, no mobile. XSaved already has iOS in beta.
- No integrations. No Notion, no Sheets, no Zotero, no RSS, no API.
- X only. If you also save on LinkedIn or Reddit, we do nothing for you.
- No MCP endpoint yet. ContextBolt has one shipping.
- The free tier has no search at all. It gives you unlimited capture, the plain text files, browsing, filtering by author, content type, date and folder, and full export — but every search path, Chinese or not, sits behind the paid licence. That is a pricing decision and we would rather state it plainly than bury it.
- We depend on X's private interfaces, exactly like every product on this page. X changes them; when it does, archiving breaks until we ship a fix. Any tool claiming this risk does not exist is not being straight with you.
What none of them do
Show you the file. Every product here says your data is “in your browser” or “on your device.” None will show you a path you can open. We save plain text files in a folder you chose, and we put a screenshot of it on the home page.
Split Chinese into words, then rank what comes back. We downloaded and unpacked all five store builds to check this properly, and the honest answer is more interesting than “nobody does it.” ContextBolt can find a Chinese phrase — its index has no Chinese in it, but when nothing matches by word its query code scans the raw text instead, so a run of characters does match. Type two words and each is matched on its own; results are ordered by how many of your words appear and then by date, with no word splitting. XSaved’s indexer replaces every Chinese character with a space, so a plain Chinese query returns your most recent bookmarks rather than the matching ones; searching inside a folder matches the raw text, and quoting the query helps only for very short phrases, because longer ones still have to pass through that same index first. Twillot’s in-app search help states that a space means or and that * stands for variable text — literal matching, which gets you nowhere in a language that does not put spaces between words. Dewey and TweetsMash search on their own servers, so we cannot check them and we do not claim to. What none of them do is split Chinese into words and rank results by how well they match.
Hand the search off between two engines so you never wait for an index. Building a real index over hundreds of thousands of posts takes the better part of a minute, and most tools would make you sit through it. Your first search is answered by a simpler engine — about 0.25 seconds on a 38,000-post archive, about 1.6 seconds at 377,000 — while the real index builds in the background; when it is ready it takes over, and from then on a search costs tens of milliseconds. The index is saved, so opening the library again reuses it — change one percent of your archive and only that one percent is rebuilt. To put a real number on it: on a genuine 38,000-post archive we measured the full index taking about 25 seconds to become ready — and you can search the whole time it builds. The 377,000 figure is a synthetic stress corpus, not an amount we expect you to have.
Sell it once. The only lifetime options in this category are $225 and $198. The $29–39 band is empty.
Two questions the table does not answer
Does this work if I have tens of thousands of bookmarks?
Yes. Archiving is incremental and idempotent — you can do it across several sessions, and re-scrolling a range you already archived costs nothing and doubles as a check for newly-deleted posts.
Is this against X’s terms?
Normal capture only reads what the page loads. You start the initial collection. Requests use that page’s own request template at a slower pace. You can stop collection at any time. MarkVault never writes to X, stores login headers or sends them to us.
Still reading?
You have just read the case against us, sourced, with our own shortcomings left in. If the one specific person it describes is you, leave an email — you will hear once, on launch day, and the founding price is limited to 100 licences.
Sources
- Twillot pricing & FAQ — 免费版 1,000 条；$2.49 / $4.99 / $9.99 月付档；「In your own browser. The index is local…」
- Twillot 中文站 — 完整中文化的定价页与 FAQ
- Dewey pricing — Free / $10 月 / $7.5 年折算 / Lifetime $225 / Export Pass $50「Valid for 48 hours」
- Dewey 首页 — 无 FAQ、无隐私或数据存储说明
- XSaved pricing — 2026-09-09 重新核实：「XSaved for Chrome is free to install and use. XSaved Pro is sold in the iPhone app.」Chrome 版标为 $0 always，含捕获、整理、检索、文件夹、笔记、标签、导出与可选同步；Pro（语义检索、Smart topics）只在 iPhone 应用内售卖。2026-08-20 首次核实时该页为 $7.99 月 / $79.99 年、X Premium 半价、「We’re considering a lifetime option…」——价格已整体撤下
- XSaved FAQ — 「Everything works locally with no account: your bookmarks are stored and searched on your device.」
- Tweetsmash pricing — Reader Pass $7/月（年付 $84）/ Yearly Pass $99 / Exports Pass $49 一次性 / Believer Pass $198 买断
- Tweetsmash FAQ — 「Once imported, your bookmarks are stored in TweetSmash.」
- ContextBolt bookmarks — Free 150 条 / Pro $6 月；FAQ：「The only thing sent to our servers is the text we process for tagging and search.」——与其首页
Local-first徽章并存
Spotted something outdated or unfair? Email support@markvault.app — this page only works if it stays true.
Writing scripts? Each line of the archive is a JSON object (JSONL). See the format documentation for its fields.