How is FSRS different from SM-2 (classic Anki)?

SM-2 uses a single ease factor that changes with each review. FSRS uses a richer three-parameter model (S, D, R) fit against hundreds of millions of real reviews. It's measurably more efficient — fewer reviews for the same retention — and it degrades more gracefully when you miss days.

Can I see the FSRS state of my cards?

Yes. Every card exposes its current S, D, and R values. Open the card inspector to see the parameters and the forecasted retention curve. Classic (open-source macOS app) lets you inspect and adjust them directly.

What retention target does Revu use?

Default is 0.9 — a card is rescheduled when we project you'd have a 90% chance of recall. You can raise it (harder, more reviews) or lower it (easier, fewer reviews) per course. We never go below 0.8 — that's diminishing returns territory.

Is the FSRS code open-source?

Yes. Revu Classic's FSRS implementation is GPL 3 licensed. You can read every line, audit the math, and verify the parameter calculations yourself.

Does FSRS work for all subjects?

It works wherever spaced recall makes sense — vocabulary, definitions, formulas, conceptual facts, procedures. For pure skill practice (coding, sport) it's less applicable. Revu automatically picks FSRS vs. alternative scheduling per content type.

FSRS Algorithm·All features

The algorithm
that remembers for you.

FSRS — the Free Spaced Repetition Scheduler — is the research-backed algorithm that powers every review in Revu. It learns from your actual recall to place each card at the precise moment before you'd forget it.

FSRS v5Open sourcePer-card inspector0.9 retention target

Start free trial See pricing

Card inspector · FSRS

"RNA polymerase binds to the…" · Biology 101

Reviewed today

Stability

16.4d

Difficulty

4.2 / 10

Retention

0.92

Interval

14 days

Retention forecast

Each review stretches the forgetting curve

R(t) = 0.9 ^{t / S}

Day 0Day 30Day 60Day 90

S = 1d

After 1st review

S = 4d

After 2nd

S = 16d

After 3rd

S = 60d

After 4th

The core loop

The algorithm in four variables.

0

Parameters per card

Stability, Difficulty, Retention — a richer model than SM-2's single ease factor.

0%

Default retention target

Adjustable per course. We don't recommend going below 0.8.

0+

Research papers

Open, peer-reviewed research underpinning the algorithm.

0%

Open source

Every line of Revu's FSRS implementation is GPL 3.

Stability

The memory strength that grows.

Stability (S) is the number of days until your recall probability drops to 90%. Every successful review stretches S — the card becomes more robust. A new card might have S = 1 day. After 5 good reviews, S could be 200 days.

Grows with success
Each passed review increases stability proportionally to current S and D.
Resets on failure
A lapse resets stability (not to zero, but heavily discounted).
Drives next interval
Interval = S · ln(0.9) / ln(0.9) — capped to your retention target.

Stability over 6 reviews

1d

4d

16d

40d

88d

160d

Review 1Review 6

Stability grows roughly geometrically — the reason intervals stretch so rapidly after a few good reviews.

Difficulty

A per-card friction coefficient.

Difficulty (D) captures how hard a card is for you, specifically. Cards with higher D grow stability slower — they need more reviews to reach the same robustness. A 'Hard' grade nudges D up; an 'Easy' grade nudges it down.

Per-card, not per-deck
Every card has its own D, fit to your behavior.
Updated on every grade
Hard → D increases. Easy → D decreases. Good holds steady.
Visible
See the D value in the card inspector and filter by it.

Difficulty distribution · 186 cards

1

2

3

4

5

6

7

8

9

10

Easy (D ≤ 3)MediumHard (D ≥ 8)

The math, open

No black box. Read the formulas.

Revu Classic's FSRS implementation is fully open-source. Below are the three core equations — the retention forecast, the stability update, and the next-interval calculation. All parameters are inspectable, all fits are reproducible.

GPL 3 licensed
Read every line. Audit the math. Fork it.
Community-vetted
FSRS has 15+ research papers and an active maintainer base.
Parameter fitting
Global defaults from a 220M-review dataset; per-user fitting optional.

Retention

R(t) = 0.9^(t/S)

Probability you'll recall the card at time t, given current stability.

Stability update

S' = S · (1 + e^a · (11 − D) · S^(−b) · (e^(c·(1−R)) − 1))

How much stability grows after a successful review. Depends on D, S, R, and your pass grade.

Next interval

I = S · ln(r) / ln(0.9)

Days until next review, derived from target retention r and current stability.

Why FSRS

The successor to SM-2.

For 30 years, SM-2 was the state of the art. FSRS is the first widely-adopted algorithm that out-performs it on real-world data — and it's open.

Fewer reviews per retention

FSRS typically needs 20–30% fewer reviews than SM-2 for the same retention target. Less time, same mastery.

Richer memory model

Three parameters (S, D, R) vs. SM-2's single ease factor. More expressive fit to real memory behavior.

Graceful degradation

FSRS handles missed days better. SM-2 tends to over-penalize — FSRS models the actual forgetting curve.

Tunable retention target

Set your own retention goal per course. 0.9 is the sweet spot; 0.95 if you're a perfectionist; 0.85 for lower-stakes content.

Inspectable state

Every card exposes S, D, and R. Open the inspector to understand exactly why a card is where it is.

Continuously improving

FSRS v5 improved on v4. v6 is in active development. Revu tracks upstream — no algorithmic dead-end.

Dig into the algorithm

Frequently asked

FSRS, answered.

FSRS (Free Spaced Repetition Scheduler) is an open-source memory algorithm developed by Jarrett Ye and a research community including members of the Anki team. It models memory using three variables per card — Stability, Difficulty, and Retention — and schedules each review at the optimal moment before you'd forget.

Keep exploring

More of Revu.

All features

Memory Analytics

Retention curves, mastery trends, time-per-topic.

Course Studio

AI generation from your lectures, slides, and notes.

Smart Scheduling

Calendar-aware plans that write themselves every night.

The algorithm that remembers for you.

Try Revu. Let FSRS decide when every card comes back. Stop guessing at intervals.

Start 7-day Pro trial Compare plans

No credit card required. Cancel anytime.