Home › Study Guides › The Best Spaced Repetition Apps, and How the Science Works
The forgetting curve, what SM-2 actually does, and eight apps ranked by who each one is really for.
A spaced repetition app does one thing a normal flashcard app does not: it decides when you see each card. Cards you know well disappear for weeks. Cards you keep missing come back tomorrow. That scheduling, rather than the flashcards themselves, is what makes the technique work, and the difference in retention is large enough that medical students have organised their entire study culture around it.
Full disclosure before anything gets ranked: we make CoStudy, one of the apps below, and we will say where the others beat us. On prices, this market changes them constantly, so exact figures appear below only where we verified them against the provider directly; elsewhere you get a description of the tier rather than a number, because a stale price is worse than none.
The science, briefly. In 1885 Hermann Ebbinghaus memorised lists of nonsense syllables and tested himself at intervals, producing the forgetting curve: memory decays fast at first, with much of new material fading within days, then more slowly. But each time you successfully recall something right as you are about to forget it, the curve flattens, and the next forgetting takes longer. Spaced repetition exploits that. Instead of reviewing everything daily, which is wasteful, or cramming once, which decays fast, you review each fact at expanding intervals of a day, three days, a week, a month. Reviews land near the edge of forgetting, where recall effort is high and therefore memory strengthening is high.
Most spaced repetition software descends from SM-2, an algorithm Piotr Wozniak published in 1987 for his SuperMemo program, and its mechanics are simple. Each card gets an ease factor starting around 2.5. After each review you grade yourself, and if you got it easily the next interval multiplies by the ease factor, so one day becomes about two and a half, then about six, then a fortnight. Fail the card and the interval resets short while the ease factor drops, so that card shows up more often from then on. Anki runs a modified SM-2 with an optional newer algorithm available, and CoStudy runs SM-2 under the hood. The exact algorithm matters less than people think; showing up for the reviews matters more. One more piece is essential: active recall. Spaced repetition schedules the reviews, but the review itself has to force retrieval, whether by flipping a card and honestly answering before looking or by answering a multiple-choice question. Rereading notes on a schedule is not spaced repetition, it is procrastination with a calendar.
Anki is the standard. Free on desktop and Android, 24.99 dollars on iOS, open source and endlessly customisable, it is the reference implementation of spaced repetition and the reason the technique went mainstream in medical schools, with add-ons covering everything from review heatmaps to image occlusion. The downsides are that it ships with zero content, the interface has not meaningfully changed in a decade, and the settings page reads like a flight manual. It is worth it if you will use it for years.
CoStudy is our app, and the pitch is simple: the same SM-2 scheduling Anki uses, but pointed at 222 pre-built question banks holding 73,642 questions across school subjects, college-entrance material and professional certifications, with practice exams and multiple-choice modes alongside. You pick a deck and start, and the scheduling runs itself. Academic access is a monthly subscription with a seven-day free trial, professional certification decks are priced individually with family bundles, and the first ten questions of every deck are free so you can check whether our content beats what you would build yourself before spending anything. Where Anki still wins: custom card creation depth, add-ons, and total control over the algorithm.
SuperMemo is the original. Wozniak's own software is still developed and its current algorithms are, by his own data, the most efficient ever built. It is also Windows-centric, visually chaotic, and takes dedication just to navigate. For algorithm purists.
Brainscape uses confidence-based repetition, where instead of pass-fail grading you rate confidence on a five-point scale and cards are spaced accordingly. The interface is the cleanest on this list and the certified decks for advanced-placement subjects, medical admissions and bar preparation are decent. It runs a monthly subscription over a thin free tier.
RemNote puts spaced repetition inside your notes: write lecture notes, flag a line, and it becomes a scheduled flashcard. For students who already take detailed notes this collapses two workflows into one. The learning curve is real but shallower than Anki's, and there is a paid pro tier over a usable free one.
Mochi is markdown-native and a quiet favourite among programmers. Cards are markdown files, the design is genuinely pleasant, and the spaced repetition is solid. There is no content library, and sync sits behind a modest subscription.
Memrise handles spaced repetition for languages, and languages only since it dropped community courses, but the spacing of vocabulary alongside native-speaker video is well executed.
Quizlet does spaced repetition, sort of. Its Learn mode has a scheduling layer, but it is shallow compared with true SM-2 systems and it sits behind Plus, which currently runs about twenty dollars a year. It is included here because people ask; if spaced repetition is the reason you are shopping, the apps above all do it better.
Three questions settle which to choose. Will you make your own cards? If yes, and making cards is itself good studying, then Anki, RemNote or Mochi. If no, you need a content library, which means CoStudy or Brainscape. What are you studying? A language points to Memrise or Anki with a frequency deck; medical school points to Anki with the large community decks that are the standard there; school subjects or a professional certification point to CoStudy or Brainscape. And how much setup will you tolerate? Be honest about this one, because the most common spaced repetition failure is not the algorithm, it is abandoning the app during week one of configuration. An app you will actually open daily beats a theoretically optimal one you will not.
Whatever you pick, the habit is the product. Ten minutes of reviews daily beats a three-hour Sunday session, because the schedule only works if you are there when a card comes due. You can feel the difference yourself without paying, since the first ten questions of any CoStudy deck are free and the scheduler starts working from question one.
A few questions come up repeatedly. On the best app overall, Anki if you are willing to build your own cards and learn its settings, since it is free on desktop and the most powerful option, while CoStudy ships the content if you want the same scheduling without building anything, and Brainscape and RemNote are strong middle grounds. On whether spaced repetition actually works, yes: the spacing effect is one of the most replicated findings in memory research, going back to Ebbinghaus in 1885, and reviewing at expanding intervals produces substantially better long-term retention than massed practice for the same total study time. On what SM-2 is, it is the scheduling algorithm Wozniak published in 1987, where each card carries an ease factor, correct answers multiply the review interval by that factor, and failures reset the interval and lower the factor. On free options, Anki is completely free on desktop and Android with the iOS app paid, CoStudy gives the first ten questions of every deck at no cost, and Mochi and RemNote have functional free tiers. On daily volume, most people sustain twenty to fifty new cards a day alongside due reviews while medical students often push past a hundred, and you should start low, because the failure mode is adding two hundred new cards in week one, facing a six-hundred-card backlog in week three, and quitting. And on whether Quizlet counts as a spaced repetition app, only loosely, since its Learn mode schedules some repetition but does not use per-card expanding intervals the way SM-2 systems do, and the feature sits behind its paid tier.
Read this in the CoStudy app →