| SM-2 | FSRS | |
|---|---|---|
| Created | 1987 | 2022-2023 |
| Based on | Limited experimental data | ML trained on 700M+ reviews |
| Approach | Fixed formula with manual tweaking | Adapts to personal memory patterns |
| Efficiency | Baseline | ~20-30% fewer reviews |
| Anki Status | Default (modified version) | Optional since Anki 23.10 |
SM-2 (SuperMemo 2)
SM-2 was created in 1987 by Piotr Wozniak. Anki uses a modified version of SM-2 that has been significantly changed from the original and is more configurable. It remains Anki's default algorithm.
How it works
- Uses an "ease factor" that adjusts based on your ratings
- Fixed intervals multiply by the ease factor
- Manual configuration of learning steps, graduating interval, easy interval, etc.
Pros
- Well-understood and battle-tested
- Highly configurable for power users
- Predictable behavior
Cons
- Based on limited 1980s experimental data
- One-size-fits-all approach
- Ease factor can spiral down ("ease hell")
- Requires manual tuning to optimize
FSRS (Free Spaced Repetition Scheduler)
FSRS was developed in 2022-2023 using modern machine learning techniques.
How it works
- Based on the "Three Component Model of Memory" (stability, difficulty, retrievability)
- Predicts probability of recall for each card
- Schedules reviews to maintain target retention rate (default 90%)
- Can be personalized by training on your review history
Pros
- More efficient: 20-30% fewer reviews for same retention
- Adapts to individual learning patterns
- Based on modern research and massive real-world data
- Less manual configuration needed
Cons
- Newer, less battle-tested
- Requires sufficient review history for optimal personalization
- Less manual control than SM-2
Why FSRS intervals differ from SM-2
FSRS tends to give longer intervals for "Easy" ratings because:
- It optimizes for efficiency — if you found something easy, you can wait longer
- It targets a specific retention rate rather than fixed multipliers
- The algorithm is more confident about scheduling based on card difficulty and your memory patterns
Our Recommendation
FSRS is recommended for most users. The Anki community consensus is that FSRS outperforms SM-2 in efficiency while maintaining or improving retention rates.
SM-2 may still be preferred if you have a heavily customized SM-2 setup that works well, need maximum manual control, or are using older Anki versions without FSRS support.
What Anki Green Uses
Anki Green uses FSRS with a native Rust implementation for maximum performance. This gives you the benefits of modern spaced repetition research without any configuration — just import your deck and start reviewing.