What Is FSRS?

FSRS, the Free Spaced Repetition Scheduler, is a modern open-source algorithm that decides when to show you each flashcard. Instead of multiplying intervals by a fixed ease factor like SM-2, it estimates how likely you are to remember each card on any given day and schedules the review for when that probability drops to a target you choose, typically 90%.

Origin

FSRS was created by Jarrett Ye, a developer and researcher who worked with the Chinese vocabulary app MaiMemo. In 2022 he and colleagues published a paper at the KDD data-mining conference describing a memory model trained on hundreds of millions of review records. He released FSRS as open source the same year, first as an Anki add-on. Anki integrated FSRS natively in version 23.10 (late 2023), and the algorithm has gone through several revisions since.

How FSRS works: the DSR model

FSRS tracks three variables for every card:

Retrievability falls over time along a forgetting curve whose speed is set by stability. Each review updates D and S. A successful recall when R is low increases stability more than one when R is high, which matches the desirable difficulties principle. A lapse lowers stability sharply but not to zero.

The model has about twenty parameters. Default values come from large public datasets, and an optimizer can refit them to your own review history so the schedule matches how your memory actually behaves.

A vocabulary example

FSRS vs SM-2 vs Leitner

Why it matters for learning words

Large vocabulary decks are where scheduling efficiency counts. With thousands of words, a small improvement in timing saves many reviews a week. FSRS also gives you an honest dial: set desired retention to 90% and it aims to show cards when you have about a 90% chance of recalling them. Raising it to 95% sharply increases workload; lowering it to 85% reduces workload at the cost of more forgotten words.

Tips and common mistakes

• Use Again and Good honestly. Treat Hard as a pass with effort, not a fail; pressing Hard when you forgot distorts the model.
• Optimize occasionally, once you have enough reviews (a thousand or more is a common guideline).
• Don't set retention too high. 97% feels safe but can multiply your daily reviews.
• Remember that scheduling is only one part. FSRS decides when; the quality of your cards and of your recall decides how much you learn. See why flashcards alone fail.

Wordie does not use FSRS; its recall mode is a simpler mastery score. If you want FSRS scheduling, you can export saved words to CSV and import them into Anki (how to import words into Anki). The forgetting curve calculator shows how stability and retention relate.

Sources: Ye, J., Su, J., & Cao, Y. (2022). A stochastic shortest path algorithm for optimizing spaced repetition scheduling. Proceedings of KDD '22.