Blog
What Is FSRS? The Spaced-Repetition Algorithm, in Plain English
FSRS is the modern scheduler behind Anki and Inku. Here is what it does, how it differs from the old SM-2 algorithm, and why you do not need to understand any of it to benefit.
If you have used a flashcard app in the last two years, you have probably met FSRS without knowing it. It is the scheduler that decides when each card comes back. It is also the reason a good app can show you far fewer cards than a bad one and still leave you remembering more.
The name sounds like a tax form. The idea behind it is simple, and worth understanding once, even if you never touch a setting again.
What the name means
FSRS stands for Free Spaced Repetition Scheduler. "Free" means open source: anyone can read the code, and it costs nothing. It was built by Jarrett Ye and a community of contributors, and it is published openly on GitHub.
"Spaced repetition" is the older idea it improves on: review material at growing intervals so you see each item just before you would have forgotten it. If that idea is new to you, start with the full explainer on spaced repetition and come back. This post is about the scheduler specifically.
The core idea: predict the day you forget
Memory fades on a curve. Right after you learn a word, recall is easy. A day later it is shakier. A week later it may be gone. The whole game of spaced repetition is to catch a card at the last useful moment, when it is hard to recall but not yet lost, because that effortful recall is what makes the memory stick.
Older schedulers guessed at that moment with a fixed formula. FSRS tries to predict it. For every card, it estimates the probability that you can still recall it today. When that probability drops to your target, the card is due. Get the timing right and you review each item roughly when you need to, and almost never when you do not.
A concrete example
Say you learn 食べる (たべる, taberu, "to eat"). FSRS does not put it on a rigid "1 day, 3 days, 7 days" ladder. It watches how you actually do. Answer it quickly and the next gap might jump to ten days. Hesitate, and the gap shrinks. The schedule bends to your memory, not the other way around.
The three things FSRS tracks
FSRS models memory with three values per card. The community calls this the DSR model.
| Value | What it means |
|---|---|
| Difficulty | How hard this particular card is for you. A confusing look-alike kanji scores higher than an easy loanword. |
| Stability | How long the memory lasts right now: the number of days until your recall chance falls to 90 percent. Each successful review grows it. |
| Retrievability | Your estimated chance of recalling the card today. This is the number that decides when the card is due. |
Stability and difficulty are updated every time you answer. Over many reviews, FSRS also tunes its own parameters to fit your memory, using your history as training data. SM-2 cannot do that.
FSRS vs SM-2: what actually changed
SM-2 is the algorithm Piotr Woźniak published for SuperMemo in the late 1980s. It powered SuperMemo, and later Anki, for decades. It is a good algorithm. FSRS is what you get when you take the same goal and add forty years of data and computing power.
| SM-2 (1987) | FSRS (2022+) | |
|---|---|---|
| Model | One fixed formula with an 'ease factor' per card | Predicts your recall probability per card |
| Adapts to you | Lightly, through ease changes | Yes; learns parameters from your history |
| Reviews needed | More, for the same retention | Fewer, for the same retention |
| You set | Intervals and ease modifiers | One number: desired retention |
| Where it runs | SuperMemo, older Anki | Anki (since 2023), Inku, others |
The practical payoff is review count. In the open benchmark the FSRS team runs across hundreds of millions of real reviews, FSRS predicts recall more accurately than SM-2. More accurate prediction means you can hit the same retention while seeing each card fewer times, which is the difference between a 20-minute review session and a 40-minute one.
What "desired retention" means
FSRS gives you exactly one dial that matters: desired retention. It is the chance you want of recalling a card when it comes up. The default is 90 percent.
Turn it up to 95 percent and you forget less, but you review more, and the extra reviews climb steeply for each point you add. Turn it down to 85 percent and you review less, but you lapse more often. Ninety percent is where most people land because it balances effort against forgetting. You do not have to take my word for it, and in Inku you do not have to set it at all.
Sources and pricing
Pricing last checked: June 2, 2026. Prices can vary by country, platform, checkout, tax, and promotion.
What this means for you
Here is the honest part. As a learner, you will not feel the algorithm. You will feel its absence of friction: the right cards show up, the session ends, and the words stay. A good scheduler is invisible.
That is also why FSRS does not belong in your hands. Anki exposes its settings, which is wonderful if you enjoy tuning and a trap if you do not. Plenty of people spend more time optimizing the deck than studying it. Inku takes the opposite stance: FSRS runs underneath, tuned for Japanese vocabulary, with no parameters to touch. If you want the full control-versus-simplicity argument, that is the heart of Inku vs Anki, and the reason behind the calmer Anki alternative.
The algorithm is the easy part. Showing up is the hard part, and no scheduler fixes that for you.
Common questions
What does FSRS stand for?+
Free Spaced Repetition Scheduler. 'Free' means open source and free to use. It was built by Jarrett Ye and the open-spaced-repetition community and is published under an open licence on GitHub.
Is FSRS better than SM-2?+
For most learners, yes, in one specific sense: it reaches the same level of retention with fewer reviews, because it predicts forgetting per card instead of applying one fixed formula. SM-2 still works and powered SuperMemo and Anki for decades. FSRS is a refinement, not a revolution.
Does Anki use FSRS?+
Anki added FSRS as an optional scheduler in version 23.10 (October 2023). It is not the historical default, so older setups still run the SM-2-based scheduler unless you turn FSRS on in the deck options.
Do I need to understand FSRS to use it?+
No. That is the point. In Inku the algorithm runs underneath and you never see a parameter. You answer whether you remembered a card; FSRS decides when you see it next.
What is 'desired retention' in FSRS?+
It is the probability that you can recall a card at the moment it comes up for review. The default is 90 percent. Higher means more reviews and stronger memory; lower means fewer reviews and more lapses. Ninety percent is the usual sweet spot.
Does FSRS work for learning Japanese specifically?+
Yes. FSRS does not care what is on the card. It works the same for a hiragana character, an N5 vocabulary word, or a kanji reading. Inku ships FSRS tuned for Japanese vocabulary so you do not have to optimize anything.
Related reading
- Spaced repetition, explained without the hype
- Inku vs Anki for Japanese
- The best Anki alternative for Japanese
- Why I stopped using Anki
Try Inku, free for 7 days, or read the full roadmap.