SM-2 vs FSRS: Which Spaced Repetition Algorithm Is Better?
FSRS predicts when you'll forget a card more accurately than SM-2, so for heavy users it can schedule the same retention with fewer reviews. SM-2 is simpler, predictable, and still works well, especially if you rate honestly. For most learners, which one you use matters less than reviewing every day.
SM-2: fixed rules
SM-2 was written by Piotr Woźniak for SuperMemo in 1987, and Anki's original scheduler is based on it. Each card has an interval and an ease; your rating multiplies the interval and nudges the ease up or down by fixed amounts. It doesn't learn from your history. A card rated Good every time goes 1 day, 2, 5, 12, 30, 75 days, whoever you are.
FSRS: a model fitted to you
FSRS (Free Spaced Repetition Scheduler) is an open-source algorithm by Jarrett Ye and the Open Spaced Repetition project, available in Anki since version 23.10. It models each card's memory with three numbers (difficulty, stability and retrievability) and fits its parameters to your own review history. You choose a target retention, such as 90%, and it schedules each card for when your predicted recall drops to that level.
What FSRS fixes
- No ease hell. FSRS has no ease factor to sink. See what is ease hell.
- Overdue cards handled sensibly. It accounts for how long it actually was since you last saw a card.
- A retention dial. You can trade more reviews for higher retention, or the reverse, explicitly.
When SM-2 is fine
- You're new, without much review history for FSRS to learn from.
- You rate honestly and use Easy when cards are easy, which keeps ease healthy.
- You want to know exactly when a card will come back.
What Decks uses
Decks uses SM-2 with Anki's standard settings and a rule that Good and Easy always add at least one day. How spaced repetition works in Decks has the details and the research behind it. If you want FSRS today, use Anki.
Published 2026-10-07.