If you already have a tutor, most language apps are the wrong tool

I've watched this go wrong the same way for years. Someone books a weekly hour with an italki tutor, has a genuinely good lesson, and then downloads Duolingo because it seems like the responsible thing to do between sessions.

Six weeks later the learner has a 42-day streak and still can't produce the subjunctive their tutor drilled in week 2.

Both things were happening. They just weren't happening to each other.

The only test that matters is whether it touches your lesson

Every roundup I read grades apps on features, price and how fun the lessons are. For a tutored learner, all of that is beside the point, because you already have the expensive part.

Here's the only test I'd apply: does this app do anything with the material from my lesson?

Whether an app teaches Spanish well is a question for someone with no teacher. You have one. Your problem is the 6 days between seeing them.

The lesson test cuts the field hard. Most of the apps on any "best language apps" list are course apps: fixed syllabus, their content, their order.

They fail the test by design. A few of them are still worth running, for reasons I'll get to.

Do you need an app if your tutor sets homework and checks it?

Probably not, and the honest version of this page has to start there, before any of the apps.

Say your tutor assigns homework and then actually checks it at the start of the next lesson. That tutor has already closed most of the gap every app on this page sells against.

The homework makes you produce the material instead of reading it, and the check creates a real deadline. The material is also the right material, which is the thing no course app can offer.

If checked homework describes your lessons, do the homework and close this tab. I'd rather say that than pretend a subscription improves a system that's working.

The gap checked homework leaves is timing, and it's narrower than app marketing suggests. Homework tends to get done in one sitting, usually the night before.

Cepeda and colleagues put numbers on the timing in 2008. Across more than 1,350 people, the gap that produced the best retention was about 20 to 40% of a 1-week test delay.

On a 7-day lesson cycle, that points at a first review 1 or 2 days after you walk out. But a week sits at the very short end of what they actually tested, so read it as a direction. It isn't a prescription.

So I'd split the homework across 3 short sittings instead of one. That number is mine, and it costs nothing.

Plenty of tutors don't set homework, or set it and never look at it again. That's when the week goes quiet, and that's the reader the rest of this page is for.

The full routine is in what to do after a language lesson.

Pimsleur fails the test and is still worth running, for the commute

Pimsleur fails my test and I'd still recommend it to a lot of tutored learners, because it does one job nothing else on this list does well.

Pimsleur is audio-first. You can run it driving, walking the dog, or doing dishes, with no screen and no hands.

If you commute, that's found time that was never going to go to your lesson notes anyway. So it isn't competing with anything.

Pimsleur's method also has real spaced-repetition ancestry, and I'll concede that plainly. Its four named principles are anticipation, graduated interval recall, core vocabulary and organic learning.

Graduated interval recall means learners "are reminded of new words at gradually increasing intervals." That's the same idea a modern scheduler automates. Pimsleur was doing it on tape decades before the apps.

Pimsleur has two tiers, Premium for one language and All Access for all 50+ languages, both with a 7-day free trial.

Pimsleur's subscriptions page shows no dollar figure. The plan data on its own product pages does, and on 21 September 2026 it listed Premium at $19.95 a month. Its blog rounds both tiers to "around $20 per month." Every Pimsleur price is here.

Pimsleur's limit is fixed content. When your tutor corrects the way you're using por and para, Pimsleur has no idea that happened and never will.

Babbel is worth it for the one grammar point that keeps eating your lessons

Babbel is the other genuine concession, for a different job: structured grammar you keep getting wrong.

Say your tutor has flagged the imperfect versus the preterite 3 lessons running. They can explain it again, but explaining it a fourth time is an expensive use of an hour with a native speaker.

Babbel has a course unit on exactly that, and its lessons "most take around 10-15 minutes." It'll walk you through it as many times as you need, without checking a clock or charging by the hour.

Babbel's Review feature does use spaced repetition, with vocabulary from completed lessons added automatically and sessions scheduled for you.

Babbel doesn't publish the interval ladder, so I can't tell you how it compares to FSRS (an open-source algorithm that predicts when you're about to forget each card). Anyone who tells you they can is guessing.

On price, Babbel offers 1, 3, 6 and 12-month plans plus lifetime. The annual works out around $9 a month.

I'd take that with salt. The figure comes from Babbel's own coupon page, so it's a discounted rate with no promise it stands, and Babbel discounts constantly.

One thing worth knowing if you found Babbel while shopping for lessons: Babbel Live is no longer available to individual consumers. If you want live teaching, that's italki, Preply or a local tutor.

Duolingo has a real published scheduler and still the weakest case here

Duolingo is the app tutored learners actually download, and it's the one with the weakest case.

Let me be fair first, because a lot of roundups I read are stale about Duolingo, and stale is how you lose an argument.

Duolingo published and open-sourced a real spaced repetition scheduler: half-life regression, in Settles and Meeder's ACL 2016 paper. That's peer-reviewed work, which is more than most competitors on this list can claim.

And the old line that Duolingo only gets you to A2 is now out of date. On 22 April 2026, Duolingo said its Spanish course teaches content through CEFR B2, across 9 courses. (CEFR is the European scale of language levels, from A1 up to C2.)

That's a claim about what the course contains. It says nothing about learner outcomes. But repeating the A2 line in 2026 is just wrong.

Duolingo is also genuinely free. The core course is there, ad-supported, gated by Energy, which replaced hearts as the free-tier limiter.

Super Duolingo removes that. It has a 14-day free trial on the US listing, and App Store prices run from $9.99 to $119.99 depending on term and plan.

Duolingo is built to be your whole course, which is exactly what you don't need. It has no mechanism for learning that your tutor spent 20 minutes on se me olvidó, so it can't put that back in front of you on Thursday.

Duolingo's streak is the sharper issue. It's very good at making you show up, and the 15 minutes it collects each night are the 15 minutes your lesson material needed.

Doing Duolingo feels like language study, so the itch gets scratched and the notebook stays shut.

If you enjoy Duolingo, keep it. Just put it in the same drawer as a crossword, and don't let it be the reason you skipped the review.

Quizlet takes your material, and may only schedule it at a desk

Quizlet at least passes my test in principle. You type your lesson vocabulary in, and it's your material.

Updated 21 September 2026. This section used to say Quizlet had no schedule at all. When I wrote it, a full-text scan of all 164 English articles in Quizlet's help centre returned zero hits for "spaced repetition".

Ten days later, on 21 August 2026, Quizlet published an article documenting one. So the old claim is dead, and this is the correction.

Quizlet's new scheduler gives you four rating buttons inside Flashcards. Depending on which you press, a card comes back "within minutes", "soon", "about a day" or "four or more days".

It still doesn't give you a named algorithm or any published parameters. A re-scan of all 165 articles on 21 September 2026 returns zero hits for "forgetting curve", "memory score", "scheduled review" and "interval".

And Quizlet's help centre says it "is currently only available on the Quizlet website". But the App Store listing for version 10.56 of the iPhone app (13 October 2026) lists Spaced Repetition as a feature, so Quizlet's own pages disagree. If your review happens on a phone between other things, check the app before you count on it. The full Quizlet comparison works through what that costs a language learner.

Beside the scheduler sits Progress, a paid feature per Quizlet's help centre, and it's worth understanding precisely. It buckets terms by how often you got them right (Not Studied, Still Learning, Mastered), with no time dimension at all.

There are no due dates and no next-review calculation. You decide when to come back.

For a tutored learner, Progress is the wrong shape, because deciding when to come back is the exact task that fails. By week 3 you've got 3 lessons of vocabulary in one place and nothing tells you which 9 items are about to go.

Worth conceding: Learn and Test are both paid for individuals. But students in a teacher-created class get unlimited ad-free Learn and Test on sets added to that class page. If your tutor runs a Quizlet class, that's real and it's free.

Quizlet's audio is also free in Flashcards mode. It's machine text-to-speech, with no native recordings, and Quizlet warns some languages have no TTS at all.

Anki is the honest recommendation, if you will actually build the cards

Anki is the honest recommendation for most tutored learners, and I'd rather send you there than oversell what I've built.

Anki ships no language content of its own. For anyone else, that's the weakness. For you, it's the whole point: an empty scheduler is what you want when the content should come from your lessons.

Building the cards yourself has its own effect, too. Slamecka and Graf named the generation effect in 1978: material you generate is remembered better than the same material read. Typing up your lesson items is itself study.

One correction to something you'll read everywhere. Anki's default scheduler is SM-2, the 1987 SuperMemo algorithm, not FSRS.

FSRS has been built in since Anki 23.10 in October 2023, and it's excellent. But you have to switch it on yourself, in the last section of the deck options screen.

SEO blogs claiming FSRS became the default in 23.10 are wrong, and it matters here. If you install Anki expecting a modern scheduler and never touch the settings, you're running a 1987 formula.

Turn FSRS on and you get FSRS-6, which fits 21 parameters to your own review history. It lets you set a desired retention target, the share of cards you want to still remember when they come due (the default is 90%). More on the mechanics in the FSRS post.

Cost: desktop is free on Windows, macOS and Linux, AnkiDroid is free, and AnkiWeb sync is free.

AnkiMobile on iOS is US$24.99 one time, which funds development. Regional prices differ.

Anki's failure mode is specific. Card-making happens after the lesson, when you're tired, and a good card takes a couple of minutes to write. Skip 2 weeks and the backlog is what kills it.

I've written up the manual system and where it breaks in Verbamor vs notes and Anki.

Verbamor costs about one lesson a month, and I have no retention number of my own

Verbamor exists for one narrow case: you keep paying for lessons and keep losing them, and you've already proved to yourself you won't build the cards by hand.

Verbamor records the lesson and picks out every word or phrase you got wrong. Those become the cards, scheduled with FSRS so each one comes back near the day it would have gone.

The material is your tutor's material. That's the entire pitch.

Now the part I'd want to read if I were you, and the part Anki wins on. Anki publishes a scheduler you can inspect: FSRS-6, 21 parameters, a 90% retention default.

I can tell you Verbamor runs FSRS too, and there my evidence stops. I have no retention number of my own to show you.

Verbamor hasn't been in enough hands for long enough to produce one that would mean anything. And a percentage invented for a comparison page is worth less than nothing.

So hold me to the bar I set below instead of a statistic. Everything measured on this page was measured by Karpicke, Cepeda and Slamecka.

It's evidence that retrieval and spacing work. It doesn't prove my app executes them well, so judge that part on the trial.

Since I've published everyone else's prices, here's mine. Explorer is $19.99 a month. Native is $129.99 a year, which works out at about $10.83 a month.

The 7-day free trial is on that annual plan. Prices are USD, and the App Store converts regionally.

The comparison I'd actually make is against the thing you're already buying. A full hour on italki commonly lands somewhere in the $10 to $30 range. Tutors set their own rates and trial lessons run cheaper, but a month of Verbamor costs roughly what one lesson costs.

That's the bar I'd hold it to. If it doesn't save you at least one lesson's worth of material a year, it isn't worth the money, and you should cancel it.

The evidence underneath Verbamor is the same evidence I'd cite for doing this on paper. Karpicke and Roediger had students learn 40 Swahili-English pairs, then varied what happened after the first correct answer.

Items that kept being tested came back at about 80% a week later. Items that were only restudied came back at about 36%.

Items that kept being tested came back at 80% a week later; items only restudied, 36% Students learned 40 Swahili–English pairs, then what happened after the first correct answer was varied: keep testing the item, or only restudy it.

80%

recalled when items kept being tested after the first correct answer

36%

recalled when items were only restudied after the first correct answer

Karpicke & Roediger 2008, Science 319(5865). Evidence that retrieval and spacing work. It doesn't measure any app on this page, mine included.

The flat 50% guess is the uncomfortable detail. The students couldn't feel which method was working. That's why "I'll just review my notes, it feels fine" survives contact with reality for so long.

Two things I won't claim. Verbamor won't make you fluent, and it does nothing about the fact that speaking is frightening. That's a separate problem with its own research and its own post on speaking anxiety.

It also won't move you off an intermediate plateau on its own. Plateaus are usually about the kind of input you're getting, more than how many words you've retained.

Nothing here is a ranking, because the answer turns on a fact about you

Nothing here is a ranking. The right answer depends on a fact about you: whether the manual version survives past week 3.

The six questions people actually ask, answered short

---

Every measured claim above traces to one of these

Prices were checked on 11 August 2026 and change often, so verify before you buy.