The FSRS Algorithm

The next generation of spaced repetition technology

What is FSRS?

FSRS (Free Spaced Repetition Scheduler) is a modern, open-source spaced repetition algorithm that represents a significant leap forward in learning technology. Unlike older algorithms that use fixed intervals, FSRS uses machine learning to understand your unique memory patterns and optimize your review schedule accordingly.

The result? 20-30% fewer reviews while maintaining the same level of knowledge retention. That means you learn faster, remember longer, and waste less time on unnecessary reviews.

🧑‍💻 The Story Behind FSRS

FSRS was developed by Jarrett Ye, a research engineer from MaiMemo Inc. and a passionate member of the Open Spaced Repetition community. Jarrett's journey with spaced repetition began in high school when he discovered Anki and experienced firsthand how powerful these tools could be for learning.

In August 2022, motivated by feedback on his academic research, Jarrett set out to create a better spaced repetition algorithm. He built upon the DSR (Difficulty, Stability, Retrievability) model proposed by Piotr Wozniak, the creator of SuperMemo, and the DHP model from MaiMemo.

The first usable version (FSRS v1) was released in October 2022. Since then, the algorithm has gone through multiple iterations, with each version bringing significant improvements. FSRS v4 was integrated into Anki in November 2023, and the latest version continues to evolve with contributions from the open-source community.

How FSRS Works

FSRS uses a sophisticated Three-Component Model of Memory (DSR Model):

📊

Difficulty (D)

Measures how inherently complex a piece of information is for you. Some concepts are naturally harder to remember than others, and FSRS accounts for this individual variation.

⏱️

Stability (S)

Represents how long a memory will last before it fades. This is measured in days until your recall probability drops from 100% to your target retention rate (typically 90%).

🎯

Retrievability (R)

The probability that you can successfully recall information at any given moment. This decreases over time based on stability and is what FSRS uses to schedule your next review.

🤖 Powered by Machine Learning

FSRS doesn't use one-size-fits-all intervals. Instead, it analyzes your complete review history and uses machine learning to calculate personalized parameters that fit your unique memory dynamics.

The algorithm continuously learns from your performance:

  • If you consistently remember certain types of cards easily, it extends the intervals
  • If you struggle with specific material, it schedules more frequent reviews
  • It adapts to changes in your learning patterns over time
  • It optimizes for your target retention rate while minimizing review time

Why FSRS Outperforms Older Algorithms

📈

Dynamic Intervals

Unlike SM-2 (used in older Anki versions) with fixed multipliers, FSRS calculates optimal intervals based on your actual performance data.

🎯

Precision Scheduling

Reviews are scheduled at the exact moment when your memory is about to fade, maximizing efficiency.

🧬

Individual Adaptation

The algorithm learns your specific memory patterns, not generic population averages.

⚡

Proven Efficiency

Studies show 20-30% reduction in review workload while maintaining or improving retention rates.

🌍 Open Source & Community-Driven

FSRS is completely open-source and developed by the Open Spaced Repetition community. This means:

  • The algorithm is transparent and can be audited by anyone
  • It's continuously improved by researchers and developers worldwide
  • It's free to use and implement in any application
  • It's backed by academic research and real-world testing

FSRS has been integrated into popular learning platforms like Anki (since version 23.10) and RemNote, and now powers Grafoxi's intelligent learning system.

For the Technically Curious

FSRS uses a power forgetting curve and sophisticated formulas to calculate difficulty and memory stability. The latest version (FSRS-6) uses 21 parameters in its calculations, all optimized through machine learning on your review history.

The algorithm is based on research published in prestigious academic venues, including:

  • "A Stochastic Shortest Path Algorithm for Optimizing Spaced Repetition Scheduling" (ACM SIGKDD Conference on Knowledge Discovery and Data Mining)
  • "Optimizing Spaced Repetition Schedule by Capturing the Dynamics of Memory" (IEEE Journals & Magazine)

You don't need to understand the mathematics to benefit from FSRS—Grafoxi handles all the complexity automatically. Just focus on learning, and let the algorithm optimize your schedule.

Experience the Future of Learning

Let FSRS optimize your learning journey with Grafoxi

Start Learning Smarter