Bunpro vs Duolingo for Japanese (2026): which one actually gets you to JLPT?

Short answer: Duolingo is better at getting you to open the app every day, and its Japanese course runs out somewhere inside N5. Bunpro is better at grammar, is built around the JLPT grammar list, and keeps working all the way to N1 — but it assumes you already have the discipline Duolingo manufactures for you. If you have never studied Japanese, Duolingo first is a defensible choice. If you have a JLPT date in your calendar, Bunpro is the one doing the work.

I build BetterLingo from Japan and I have paid for both of these. This page names what each one is better at rather than pretending one wins, and it says plainly what BetterLingo is not, near the bottom.

First: Bunpro is not Bunpo

Two different apps, one letter apart, and search results mix them constantly. Bunpro is a browser-first JLPT grammar SRS with a large written explanation library. Bunpo is a mobile lesson app much closer in shape to Duolingo. Most "Bunpo vs Duolingo" comparisons online are about the other app. This page is about Bunpro.

What each one actually is

Duolingo is a gamified course. Word tiles, streaks, hearts, leagues, a daily push notification, and a path that decides the order for you. The Japanese tree teaches kana and then walks through basic sentence patterns. Grammar is taught implicitly: you meet a pattern in a sentence, you drill the sentence, and the explanation is a short tip you can skip.

Bunpro is not a course. It is a spaced-repetition system whose deck is the JLPT grammar list, N5 up to N1, with a written explanation, several example sentences and links to outside resources for every point. Reviews are cloze sentences: you read a Japanese sentence with a gap and you type the missing grammar in Japanese. It expects you to already know vocabulary and kana, and it will not teach you either.

Where Duolingo is genuinely better

  • Starting from absolute zero. Bunpro's N5 deck assumes you can read kana and know what a particle is. Duolingo does not assume anything, and the first two weeks of Japanese are exactly where most people quit.
  • Consistency. This is not a joke feature. The streak, the notifications and the five-minute lesson size are the product, and they work. An SRS you open three times a month is worth nothing, and Duolingo is measurably better at making people open an app.
  • Listening and speaking, a bit. Every sentence has audio and some exercises ask you to speak. Bunpro's audio is on example sentences and is not the point of the product.
  • Price at the free tier. Duolingo's free tier is a real, usable product with ads. Bunpro is a subscription after its trial.

Where Bunpro is genuinely better

  • Grammar explained, not guessed. Japanese grammar does not survive being inferred from word tiles. Why and not , what the te-form is doing in a given sentence, how ようだ differs from らしい — these need prose, and Bunpro writes the prose.
  • It maps onto the actual exam. The deck is the JLPT grammar list. If you are sitting N3 in December, you can see exactly which points you have not met.
  • Ceiling. Duolingo's Japanese course tops out around N5, and past that it is repeating things you know. Bunpro has N2 and N1 grammar and gets more useful as you climb, not less.
  • You produce, not select. You type the answer in Japanese instead of picking from four tiles. Recognising the right tile is a much easier task than producing the form, and the exam and real life both ask for the harder one.

The gap neither one closes

Both tools grade against an answer key. Bunpro's cloze reviews accept a small set of correct fills for one gap; Duolingo accepts a set of accepted translations for one sentence. Neither can look at a sentence you invented, from scratch, and tell you it is grammatical but unnatural, or that a native speaker would use a different particle here, or that your ので should have been から in that register.

That is the gap that shows up the first time you write an email in Japanese, and it is why most people who finish an SRS deck still feel like they cannot write. Free-form output with feedback is a genuinely different exercise from filling a gap, and it is the part almost nobody has automated.

So which do you pick?

  • Never studied Japanese, no exam date: Duolingo, for a month or two, to get kana and a habit. Then move.
  • Can read kana, have a JLPT level in mind: Bunpro. This is the case it is built for.
  • Already at N3 or above: Duolingo has nothing left for you. Bunpro plus reading is the standard stack.
  • Stalled and cannot say why: it is usually output. You can recognise far more than you can produce, and no amount of extra recognition drilling fixes that.

Plenty of people run both, and stacking two or three tools is the norm rather than a failure.

Where BetterLingo fits, and what it is not

I build BetterLingo, a JLPT practice site that marks written answers. You answer a question in your own Japanese, in full sentences, and it comes back marked with what was wrong and why — the exercise both tools above deliberately skip. It covers N5 through N1.

What it is not, plainly: it is small and new. There is no mature community-built deck library like Bunpro's, no decade of forum threads, no course path to walk from zero, no mobile app, and no gamified streak machine to make you show up. If you need to be taught kana, it is the wrong tool. If you want the deepest written grammar library in the hobby, that is Bunpro and it is not close. BetterLingo does one thing the others do not do at all, and it does not pretend to do the rest yet.

Related reading: Bunpro alternatives, Bunpro vs LingoDeer, and how to check your Japanese grammar.

Next steps

Pick one of the two above on the honest reading of where you actually are, and stick with it for a month. If the thing that keeps failing is writing your own sentences rather than recognising other people's, that is the specific problem BetterLingo was built for.