Obsidian vs Logseq: Pages or bullets on disk (2026)

This is the difference people discover after they commit, so take it first. An Obsidian note is markdown as you would write it by hand. Headings are headings, paragraphs are paragraphs, and nothing in the file exists for Obsidian's benefit. Open it in TextEdit and it reads as prose.

A Logseq page is markdown too, and it carries the outliner in its syntax. Nearly every line begins with a dash because every line is a block. Page and block properties sit inline as double-colon pairs. Block references appear as raw UUIDs in parentheses, which are meaningless in any other editor. Logseq did not hide this and it is not a bug, but you should know the shape before you have four thousand pages in it.

The Logseq database version changes the picture again by moving notes into SQLite, and two-way sync between markdown files and that database is still in development. If plain files were your reason for choosing Logseq, that roadmap is worth watching before you migrate.

Here Logseq is simply ahead. Every block has an identity, so one bullet can be embedded in five other pages and stays live in all of them. Queries pull blocks by property, tag, or page across the whole graph. Journals are the front door, so you write into today and organise later. Flashcards with spaced repetition and a task system are in the core app rather than in plugins that might stop being maintained.

Obsidian works at note granularity. You link to a note, or to a heading inside one, and you embed a whole note or a section. Dataview, a community plugin, queries frontmatter and renders tables, which is powerful and is not the same thing as addressing an individual line. If your system depends on block transclusion, Obsidian will feel like a step backwards no matter how many plugins you install.

The counterargument is that block addressing puts UUIDs in your files and asks you to think in trees. Plenty of people tried Logseq, liked the queries, and left because every thought had to become a bullet first.

Obsidian has more than 2,000 community plugins and a large active community that answers questions the same day. Plugin quality varies and mobile compatibility is a recurring problem, and the breadth is still the biggest single reason people pick it. The app has also been free for personal and commercial use since February 2025.

Logseq's development pace is its documented weak point. The database version rollout has been delayed repeatedly and its own forum carries the frustration about it. The plugin ecosystem is much smaller. What Logseq has instead is AGPL-3.0 source you can read, fork, and keep running without anyone's permission, which for some people settles the question before any feature comparison starts.

Both apps are Electron, so both ship a Chromium instance and a JavaScript runtime per window, and neither vendor publishes a benchmark you could check. What exists instead is documented community experience, and it points the same way on each side.

Logseq's community reports performance falling off past roughly 2,000 pages in the file-based version, and its phone app lags behind the desktop one. Obsidian's forum has threads about slow startup on large vaults with many plugins loaded, and its mobile app is described as slow to load big vaults. So Logseq's ceiling is about graph size and Obsidian's is about plugin count, and both bite in the low thousands of notes.

Copy two thousand of your real notes into each one and time the cold start yourself. Reputation is a poor proxy for your own hardware and your own vault.

Neither app gives you free reliable multi-device sync, and this is the shared weak spot of the local-markdown camp. Obsidian Sync is $4-5 a month with end-to-end encryption and works well. Going free means iCloud, Syncthing, or the community git plugin, whose known faults are index.lock errors, line-ending conflicts, and merge handling on mobile its users call unstable.

Logseq Sync sits in beta behind a sponsorship gate, so most multi-device Logseq users build their own with git, iCloud, Dropbox, or Syncthing. Conflicts are the known cost of that route and Logseq's own forum discusses it at length. The database version raises a harder version of the question, since merging a SQLite file has nothing in common with merging text.

Obsidian at least sells you a working answer. Logseq asks you to be your own sysadmin or wait for the beta.

Logseq is free and AGPL-3.0 with no feature held back, and its one paid piece is the Sync beta offered to sponsors. Obsidian is free too, including commercial use since February 2025, with an optional $50 per user per year commercial licence and Sync at $4-5 a month plus Publish at $8-10. Sync yourself on either app and both cost nothing. Pay for Obsidian Sync and it is $48-60 a year, which is the price of not maintaining your own git or Syncthing setup. Valutorio is sold on the Mac App Store and bills nothing for sync, given that the remote is your own. Logseq wins on price and on licence, and the gap is small unless you value the source code.