Apps like Duolingo, and the prices I couldn't verify

Three pages rank above this one for "apps like duolingo". I fetched all three today.

One lists Memrise at "$59.99/month" inside its flagship comparison table. That's the annual price wearing the wrong unit, an order-of-magnitude error sitting in the exact cell a buyer reads first, and it has been there since October 2024. Another prices only 2 of its 10 ranked apps, misspells Babbel as "Babble" in its own ranked list, and carries commercial links with no disclosure. The third ranks its own app first, gives itself the trophy in its own table, writes the evaluation rubric around its own feature set, and is the one entry with no hard price while all 11 rivals get a number.

So here's the deal I'll make instead. I'll print a price only where I could pull it off the company's own page today, and where I couldn't, I'll name the app and say the fetch failed.

That leaves gaps. The gaps are the honest part.

Start here: what did Duolingo stop doing for you

People leave Duolingo for about four different reasons and the right answer is different for each. Find yours and skip the rest of the page.

• You want the grammar explained. Babbel. It's a taught course with rules written down, roughly 10 to 15 minutes a lesson, and it's the only app here whose price I could read cleanly.
• You finished the course and forget everything anyway. A spaced repetition tool. Anki is free on desktop and Android and it's the right answer for most people reading this sentence.
• You want to speak. Buy an hour with a person. italki's community tutors typically charge $4 to $20 a lesson by italki's own published figure. No app on this page is a substitute, mine included.
• You have no hands and no screen, just a commute. Pimsleur is audio-first and built for exactly that. Its own subscriptions page publishes no dollar figure, which I'll come back to.

Notice that two of those four send you somewhere free, and one sends you to a human being.

The prices I could confirm, and the ones I couldn't

Every figure in this table came off the company's own page or its App Store listing on 21 September 2026. A blank cell means the fetch returned no readable price, and I'd rather show you the blank than guess.

That last row is the interesting one, so here's what actually happened when I tried.

Memrise's plans URL returns a 404 to a direct fetch. Busuu's premium page serves HTML containing the word "price" 19 times and not one dollar figure, because the numbers are drawn in by script after the page loads. LingoDeer's table rendered on one fetch out of four and came back empty on the other three, behind a countdown banner advertising 75% off, which tells you the standing price and the sale price are hard to tell apart even when you can see them.

None of that makes those apps bad. It makes their prices unquotable, and a roundup that prints them anyway is reading them off somebody else's blog. Which is, I think, exactly what happened to the page that has Memrise at $59.99 a month.

The question the whole category dodges

Across roughly 14,000 words on the three pages ranking above this one, there is one reference to the CEFR and nothing at all on how long any of this takes.

That's the actual purchase question. Someone leaving Duolingo is usually leaving because a 400-day streak did not produce a conversation.

The honest anchor is the US State Department's own figure for its diplomats. The Foreign Service Institute puts Category I languages at 24 to 30 weeks, or 552 to 690 class hours, with Spanish and French both at 30 weeks, to reach an integrated speaking-and-listening score of 3. That page also notes a typical week is 23 hours in class plus 17 of self-study, so 30 weeks is closer to 1,200 total hours than 690. We walk through that arithmetic in the FSI recalculation, including why nearly every blog still quotes the retired 600-to-750 figure.

Hold any $10-a-month app against that number and the question changes shape. You're not picking a subscription for this month. You're picking the thing you'll still be opening in year two.

On outcomes specifically, be careful what you accept. Duolingo says its Spanish course teaches content through CEFR B2 as of April 2026, which is a claim about what the course contains rather than about what learners reach. The strongest published result on app learning, Jiang and colleagues in CALICO Journal (2024), tested 245 learners and found them scoring Intermediate High, in reading and listening. Speaking and writing weren't measured, and three of the four authors work at Duolingo. Tap-to-translate apps train recognition well, and recognition is not production. That gap is why a long streak and no conversation go together so often.

The apps, with a named weakness each

I have not used most of these. What follows comes from each company's own documentation, read today, plus published research where it exists. Nobody here paid me and there are no affiliate links on this page.

Anki. Free almost everywhere, close to two decades old, and the best review engine in this category. Its named weakness is precise: it ships no language content whatsoever, so every card is one you typed. Migaku, which sells a competing product, puts card building at roughly 30 seconds each and about 12.5 hours for 1,500 cards. That build cost is where people quit. And the scheduler you get by default is SM-2 from 1987; Anki's manual calls FSRS "an alternative to Anki's legacy SuperMemo 2 (SM-2) algorithm" and tells you to switch it on yourself. Blogs claiming FSRS is the default since 23.10 are wrong.

Babbel. Real lessons with grammar explained, all 14 languages included in one subscription, and audio recorded by native speakers across what Babbel says is 100,000 hours of content. It beats me outright on that axis. Its named weakness comes from Babbel's own help centre: you cannot add your own words to review. Verbatim, "the vocabulary added to your review is predefined based on the lessons you complete. It is not possible to manually add random words or phrases to your review." So the word your tutor corrected on Tuesday has nowhere to go. Its review ladder is also fixed at six steps, 1 day, 4, 7, 14, 60, then six months, with no per-card adaptation.

Duolingo. The free tier is genuinely free and genuinely good, which most roundups underplay while trying to sell you something. It published a real spaced repetition scheduler, half-life regression (Settles and Meeder, ACL 2016), and open-sourced it, though whether that's what runs today isn't documented anywhere Duolingo owns. It also teaches Chess, Math and Music, so "languages only" is wrong. Named weakness: the exercises put the target sentence on screen before you speak it, so the retrieval step is done for you.

italki. Not an app in the sense the other entries are, and that's the point. Tutors set their own prices, there's no subscription, and its Lesson Summary tool writes up topics, vocabulary and grammar after a lesson taken in italki Classroom. Named weakness, and it's a sharp one: error corrections and idioms are behind italki Plus. The two most valuable things in any lesson sit in the paid tier. There are also four conditions that must all hold or no summary generates at all, including both you and the teacher enabling audio recording.

Pimsleur. Audio-first, four named method principles including Graduated Interval Recall, which is real spaced repetition ancestry and worth conceding. Best commute tool here. Named weakness: its subscriptions page names no price, and the famous "30-minute lessons" figure does not appear on Pimsleur's own method page despite being repeated everywhere.

When the whole category is wrong for you

Worth saying plainly, because it's the case a page selling apps never raises.

If your problem is that you've studied Spanish for two years and freeze when a waiter speaks to you, no subscription on this page fixes that. You need hours of talking with a person who corrects you. Buy the tutor first and add an app later, or don't add one at all.

If your problem is that you can't get through 10 minutes a day, a cheaper app won't fix it either. That's a habit question, and it's usually about when you study rather than what you study.

And if you want to build your own cards and you'll genuinely do it, use Anki and stop reading roundups. It's free, it's better than what I sell on scheduling flexibility, and the money you save buys tutor hours that will do more for you than any of this.

Here's mine, in one paragraph. Verbamor records a tutor lesson, pulls the words and phrases you actually fumbled out of the transcript, and builds fill-in-the-blank cards with generated audio and an image, scheduled by FSRS. You can also just type a phrase into the Translate tab and keep it, no lesson needed. It's iOS only, no Android and no web, and it covers Spanish, French, Italian, Portuguese and Russian. The audio is synthesised text to speech with a native-accent instruction, so Babbel's recorded human voices are better on that specific axis. $19.99 a month or $129.99 a year, 7-day trial on the annual plan, no free tier. It exists for one failure mode: you pay for lessons and lose most of what happens in them by the weekend.

I have no retention study of my own to show you, which is worth stating on a page criticising other people's evidence. What I can point at is the research page, where every claim the app makes is traced to its paper.

Questions people actually ask

---

Sources

Every load-bearing claim Verbamor makes is traced to its paper on the research page.