Best Flashcard Apps for Language Learning in 2026

For most language learners the strongest flashcard app is Anki: free on desktop, Android and the web, with speech built in, a card type made for sentences, and a review algorithm you are allowed to retune. Clozemaster is the better install if you would rather be handed sentences than write them, and Memrise if a real human voice matters more to you than card control. Quizlet gates several of the pieces that matter most here.

Almost every flashcard roundup is written for somebody revising a subject in their own language, where a card is a term on one side and a definition on the other and the only live question is which app looks nicer. A language learner has a harder brief. The card has to be pronounceable. It may not be written in the Latin alphabet at all. A bare word pair teaches something subtly wrong. And the material has to keep coming back for years rather than until Thursday. Five requirements fall out of that, and the sections below take them one at a time, naming the tier each capability sits on as they go, because on this particular query the paywall usually lands on exactly the feature you came for.

What does language learning ask of a flashcard app that revision does not?

Five things, and no app on this page hands you all five for nothing.

• Sound. A written card can teach you what a word means and cannot teach you how it is said. Either the app speaks the card, or you supply a recording, or the pronunciation you rehearse for six months is the one you invented in your head.
• A writing system. Learning Japanese, Chinese, Arabic or Russian means typing in a script your keyboard does not have, and it means the card rendering scripts correctly, including furigana above kanji and text that runs right to left.
• Sentences. Vocabulary lives inside grammar. A word paired with one English gloss trains you to translate, which is the skill you want least.
• Decks somebody already built. Nobody wants to hand-type the two thousand most frequent words of a language before studying any of them, so the question is whether a frequency list is available and what it costs.
• A scheduler you can move. Vocabulary forgets differently from exam facts, and a learner with three hundred new words a week needs a different workload from one revising twenty diagrams before a test.

None of this replaces the surrounding work, and the app is the smaller half of it. Our guide to studying a foreign language covers the part no deck does for you.

Sound is the requirement that removes most apps

Anki speaks, and it does so by borrowing a voice rather than shipping one. Putting a text-to-speech tag in a card template makes it read a field aloud, and the manual is specific about where the voice comes from: "On Windows, macOS, and iOS, Anki will use the OS’s built-in voices. On Linux, no voices are built in, but voices can be provided by add-ons". The same page states the version floor, that the feature "requires Anki 2.1.20, AnkiMobile 2.0.56 or AnkiDroid 2.17". There is no charge for any of it, and there is also no promise about quality: a language your machine has no voice for will simply stay silent, and a Linux desktop starts with nothing.

Clozemaster is honest about the same dependency in almost the same words. Its FAQ says "The text-to-speech audio depends on your browser, device and operating system at the moment", and adds that it works best in Chrome and Safari on desktop and in the mobile apps while online. Listening as a study mode of its own is a paid line: the Clozemaster Pro page sells "unlimited listening practice for over 30 languages" and footnotes it, "Only available for Fast Track, Most Common Words, and Grammar Challenge collections."

Quizlet is where the plan boundary gets sharp, and it is worth reading before you commit a term to it. Its help centre publishes the list of languages its text-to-speech covers and marks ten of them with an asterisk, footnoted "Only available for Quizlet Plus and Quizlet Plus for teachers subscribers". The starred ten are Bengali, Czech, Danish, Estonian, Filipino, Hebrew, Hindi, Hungarian, Norwegian and Vietnamese, so whether Quizlet talks to you for free depends entirely on which language you picked. The same page carries a limit that catches sentence cards later on: "If your text is longer than 300 characters, TTS audio doesn't play."

Recording a human voice is a separate gate again, and three apps put it behind money. Quizlet states it plainly: "If you have a Quizlet Plus or Quizlet Plus for teachers subscription, you can add custom audio recordings to your sets on the website." Note the second half of that sentence as well, since the feature is web-only and the article repeats that recording "is currently only available on the website". Brainscape lists "Add images and sounds to flashcards" among the things that arrive with Pro rather than with the free Basic tier, per its pricing page. Anki is the exception: adding your own audio file to a note costs nothing, and its homepage lists "Audio, images, videos, and scientific markup can be used in your flashcards" as a plain capability of the free program.

Memrise answers the question from a different direction by not synthesising at all. Its App Store listing says the app is built on filmed speakers: "Unlike gamified apps with robotic voices, it's built around real native-speaker videos". That is the best answer to the pronunciation problem on this page, and the same listing prints the gate directly underneath: Pro is what will "Unlock all native speaker videos", while the free plan is described as "Limited vocab lessons", "Limited videos and conversations" and "Ads".

Scripts, and where the typing problem actually lives

If you are learning a language written in another alphabet, the first surprise is that the app is rarely the obstacle. Anki says so in its own manual: "All modern computers have built-in support for typing accents and non-Latin characters, and multiple ways to go about it", and the route it recommends is "using a keyboard layout for the language you want to learn", noting that Japanese, Chinese, Thai and similar languages each have layouts of their own. Install the layout once and every app on the machine can accept the script, including the ones on this page that never mention it.

What differs is what happens after the characters land in the field, and Anki is alone here in documenting the annotations a script needs. Its card-template reference explains that "These annotations are known as ruby characters" and that "In Japanese, these are known as furigana", then gives the template filter that places them above the right character. None of that is metered, and it is part of the same free program.

Anki also publishes its own weak spot, which no marketing page would. The editing page warns that the toolkit it is built on "has trouble dealing with a few input methods, such as holding down keys to select accented characters on macOS", a small and real annoyance for French and Spanish learners on a Mac. The same page hands right-to-left learners off rather than solving it for them: "If you are learning a right-to-left language, there are lots of other things to consider."

The other three document sound rather than script, so judge them on that. Quizlet’s spoken-language list, on the same audio page as before, covers Japanese, Korean, Mandarin Chinese, Arabic, Russian and Hebrew, and Hebrew is one of the ten that carry the subscriber asterisk. Clozemaster prints its own supported list on the Pro page, reaching Japanese, Korean, Mandarin Chinese, Arabic and Hindi, gated by the same rule as the rest of listening. Memrise splits the writing system into separate courses at enrolment, its App Store listing naming both "Japanese" and "Japanese (no script)" among the courses, on the free plan that its own comparison calls "Limited vocab lessons".

A word pair is the wrong unit

This is the section that most separates language study from revision, and it is where a general roundup stops being useful. A card reading abandonar on the front and to abandon on the back trains a lookup. What you want trained is the word arriving inside a sentence, conjugated, in an order that sounds like the language.

Anki ships the mechanism for that free, in the cloze note type. The manual defines it as it works: "Cloze deletion is the process of hiding one or more words in a sentence", and it adds that "Anki provides a special cloze deletion type of note, to make creating clozes easy". Write one sentence, hide the verb, and you have a card that tests recall of a form in context. Hide a second word and Anki makes a second card from the same sentence at no extra effort.

Clozemaster is that idea turned into an entire product. Its homepage describes "Thousands of fill-in-the-blank sentences in 50+ languages", and its FAQ explains that "Clozemaster is a game where you improve your language skills by filling in the missing word for thousands of sentences". The sentences come from the Tatoeba corpus and the ordering comes from frequency data, the same FAQ stating that "The frequency lists used are from Wiktionary" apart from Japanese, which is ordered by JLPT level. That combination, a real sentence with one word removed and the whole set ordered by how common the word is, is the closest thing here to a purpose-built vocabulary curriculum.

Quizlet will hold a sentence in a term field, and the 300-character ceiling on its text-to-speech is where that quietly stops working: pass it and the card goes silent with no warning attached. If you want the reasoning behind sentence-level cards rather than the mechanics, our piece on memorising vocabulary makes the case at length.

What do ready-made decks actually cost?

Anki gives them away, and then tells you not to trust them. Downloading a shared deck is described as "The easiest way to get started with Anki", and the browsing and importing cost nothing. Two paragraphs later the same manual says something no vendor with a deck store would print: "If you are a language learner, you may be tempted to download a long list of words and their translations, but this won’t teach you a language any more than memorizing scientific equations will teach you astrophysics." Its own recommendation is that "Creating your own deck is the most effective way to learn a complex subject". Read that as the free option coming with a warning label attached by the people who wrote it.

Brainscape meters its library instead. The free Basic tier is described as "Free forever" and promises that "Creating, sharing, and studying your own content is always FREE", with "Study 2,500 pre-loaded flashcards" listed as the ceiling on other people’s material. Moving to "Study unlimited Certified flashcards" is a Pro line, alongside the sounds mentioned earlier, and the pricing page headlines Pro at 7.99 dollars a month, which its own plan data prices as 95.88 dollars billed every twelve months. Shorter terms cost more per month, and a one-time lifetime option sits on the same page at 199.99 dollars.

Memrise gates its courses the same way, its store listing giving the free plan "Limited vocab lessons" and reserving the rest for Pro. That same listing prices the monthly plan at 24.99 dollars and the annual one at 61.99 dollars, sitting alongside a row of promotional subscription offers at other figures, so treat those two as the headline rates and read the checkout for whichever offer you are shown. Clozemaster is the outlier, its FAQ answering the cost question with "Nada. It's free to sign up and play, and always will be", while the homepage adds that "Pro unlocks advanced features for power users". So the frequency-ordered sentences themselves are free, and the controls over them are not. Two figures are missing from this page on purpose. Clozemaster names no Pro price on its own Pro page, and Quizlet serves its upgrade page only to a real browser session, so both are worth reading at the checkout rather than taking from a roundup that guessed.

Quizlet brings the deepest bench of ready-made sets of anything here, and its adaptive study mode is the thing you would want to point at it. That mode is a subscription feature: "Learn is available to Quizlet Plus and Quizlet Plus for teachers subscribers", with a taster for everyone else, since "If you're not a Quizlet subscriber, you can start a free study session in Learn to see what it's like to practice with different question types." There is one genuinely free route through it that almost nobody mentions, and it needs a teacher: Quizlet says "Students in your class get unlimited, ad-free access to Learn and Test modes" when they are logged in, added to a class, and studying sets added from the class page. If you are in a taught course, ask.

Can you change how often a word comes back?

Anki is the only app here that hands you the algorithm. Its deck options page opens by saying they "primarily control the way Anki schedules cards", and the modern scheduler is described there as an alternative to the old SuperMemo 2 approach, with a dial attached: "Desired retention controls how likely you are to remember cards when they are scheduled for a review", defaulting to 90 percent. Raising it means more reviews a day and fewer forgotten words, which is precisely the trade a learner with a fixed daily hour needs to make deliberately. The manual also warns against pushing it too far, recommending "you keep it lower than 97%". All of that is free, on every platform except the one-off purchase on iOS.

Clozemaster publishes its intervals and then charges for the ability to move them. The FAQ states that "Default intervals are at 1 day, 10 days, 30 days, and 180 days based on how many times a sentence is answered correctly in a row (Pro subscribers can customize these intervals)". Pro also decides how many reviews get mixed into a round, the same FAQ noting that subscribers "can change this number". Knowing the ladder in advance is genuinely useful, and four fixed steps is a coarser instrument than Anki gives you for nothing.

Quizlet added a rating-based review that behaves like a scheduler, with four buttons and a daily queue, and two limits are stated on the page describing it. It is desktop-bound, "Spaced repetition is currently only available on the Quizlet website", which rules out the commute that is most people’s actual study slot. And the only control it names is volume rather than timing: "Enter a number in New cards per day to change how many new cards you review each day." Worth knowing too that Quizlet describes its mid tier as one that "provides extended access to premium study tools with monthly usage limits", so paying is not the same as removing the meter. Our head-to-head on Quizlet against Anki works through that pairing for study generally.

Brainscape schedules by asking how confident you felt, rating one to five, and its pricing page sells "Learn faster with spaced repetition" on the free tier. No interval control appears anywhere on that page, which is the design rather than an omission: the product exists to spare you the tuning that Anki hands over.

Read the GeniusPal price list before the feature list

Money first, because on this query it settles the question faster than any feature does. Two study-set generations come with a new GeniusPal account. That pair is counted a single time across the whole life of the account and never replenished, and both of them return as a scored quiz. On the Student plan, GeniusPal unlocks flashcards plus written active recall in one go; that plan costs 14.99 dollars each month and lifts the generation ceiling to a hundred. Genius is a single yearly charge of 59.99 dollars, and its unlimited label covers a large, finite fair-use number. So a language learner who never pays gets no flashcards from GeniusPal at all, and for vocabulary that is the format everything above has been about.

The absences matter more than the price, and they cover most of this page. There is no audio, no text-to-speech and no pronunciation of any kind, so nothing it produces can be heard. There is no language-pair setting, no ready-made deck in any language, and no filmed speaker. Spacing is absent as well: a weak pile collects whatever you fumble and the next round leans toward it, and that is the whole mechanism, with no due dates, no daily queue and no arithmetic deciding when a word returns. Anyone whose problem is any of those should pick one of the three apps higher up this page and stop here.

Its actual job is to take material you already hold and turn it into written practice questions. A grammar chapter you scanned, a vocabulary list you typed out for a class, a handout the department published: upload one and questions come back bounded by what is printed on those pages, and your answers get marked rather than self-rated. For grammar rules, case endings, set phrases and the meanings you are expected to reproduce in an exam, that is a reasonable check. For pronunciation it is no check at all, which is why it belongs after the apps above rather than instead of them.

One caution applies to every generated card, in any tool here. A question written by a model can be confidently wrong about a gender, a register or an idiom, and a language learner is by definition the person least equipped to spot it. Read the first set closely before you rehearse anything in it; our piece on how far an AI-written card can be trusted sets out what to look for.

Where to start, and the test to run in week one

Install Anki first, even if you end up somewhere else. Everything on this page is in some sense a reaction to it, and the alternatives make far more sense once you have met the thing they are reacting against. It costs nothing on a computer or an Android phone, it will speak your cards through voices your machine already has, and it is the only option here that lets you change how often a word returns without paying for the privilege. If you are on an iPhone with no computer, price that in honestly: AnkiMobile is a single purchase listed at 24.99 dollars on the US App Store, with no subscription behind it.

Then, before you build four hundred cards, run the one test that decides whether any of this works. Make five cards from the material you are actually studying, in the language you are actually learning, and play them. If the voice is one you would be willing to imitate, and the card shows you a sentence rather than a word, and the same card comes back on a day you did not choose, the system is doing its job. If any of those three fails, changing app now costs an afternoon, and changing it in March costs a year of half-learned pronunciation. For the wider view across every free option rather than the language-specific ones, our roundup of free flashcard apps covers the rest of the field.