FSRS (Free Spaced Repetition Scheduler)

FSRS, or the Free Spaced Repetition Scheduler, is a next-generation scheduling algorithm developed by Jarrett Ye and the open-source community. It replaces older algorithms like SM-2 (used by Anki since 2006) with a more accurate model of human memory. Where SM-2 uses simple heuristic rules to adjust intervals, FSRS uses a three-parameter model — stability, difficulty, and retrievability — that has been validated against millions of real review logs and consistently outperforms SM-2 in predicting when a learner will forget.

The key differences from SM-2 are significant. SM-2 treats all learners and all cards with roughly the same fixed multipliers, while FSRS adapts its parameters to each card individually. SM-2 has no concept of a target retention rate, while FSRS lets you set a desired recall probability (e.g., 90%) and mathematically computes the optimal interval to achieve it. FSRS also accounts for the "short-term memory" effect — the tendency to remember something immediately after studying it even without real learning — which SM-2 ignores entirely.

Revu uses FSRS as its core scheduling engine. Each card maintains its own stability and difficulty values that update with every review. The algorithm tracks four key parameters: stability (how durable the memory is), difficulty (how hard the card is for you), retrievability (the predicted probability of recall at any given moment), and the desired retention target. This means your review schedule is personalised not just to you as a learner, but to each individual piece of knowledge you are trying to retain.