Anki (software)

Anki

Anki 2.1.6
Developer(s)	Damien Elmes
Initial release	5 October 2006; 18 years ago (2006-10-05)
Stable release	24.06.3 / 28 June 2024; 4 months ago (2024-06-28)[1]
Repository	github.com/ankitects/anki
Written in	Python, Rust
Operating system	Windows, macOS, Linux, FreeBSD; Android and iOS (special versions)
Available in	48 (desktop)/14 (AnkiMobile) languages
List of languages Desktop: Afrikaans, Arabic, Armenian, Basque, Bulgarian, Catalan, Simplified Chinese, Traditional Chinese, Croatian, Czech, Danish, Dutch, English (United Kingdom), English (United States), Esperanto, Estonian, Finnish, French, Galician, German, Greek, Hebrew, Hungarian, Irish, Italian, Japanese, Korean, Latin, Lojban, Malay, Mongolian, Norwegian, Occitan, Persian, Polish, Portuguese, Brazilian Portuguese, Romanian, Russian, Serbian, Slovak, Slovenian, Spanish, Swedish, Thai, Turkish, Ukrainian, Vietnamese AnkiMobile: English, Arabic, French, German, Irish, Japanese, Persian, Polish, Portuguese, Russian, Simplified Chinese, Spanish, Traditional Chinese, Turkish, Ukrainian
Type	Flashcard spaced repetition
License	Desktop: GNU AGPL v3+[2] Android: GNU GPL v3[3]
Website	https://apps.ankiweb.net/

Anki (US: /ˈɑːŋki/, UK: /ˈæŋki/; Japanese: [aŋki]) is a free and open-source flashcard program. It uses techniques from cognitive science such as active recall testing and spaced repetition to aid the user in memorization.^[4][5] The name comes from the Japanese word for "memorization" (暗記).^[6]

The SM-2 algorithm, created for SuperMemo in the late 1980s, has historically formed the basis of the spaced repetition methods employed in the program. Anki's implementation of the algorithm has been modified to allow priorities on cards and to show flashcards in order of their urgency. Anki 23.10+ also has a native implementation of the Free Spaced Repetition Scheduler (FSRS) algorithm, which allows for more optimal spacing of card repetitions.^[7]

Anki is content-agnostic, the cards are presented using HTML and may include text, images, sounds, videos,^[8] and LaTeX equations. The decks of cards, along with the user's statistics, are stored in the open SQLite format.

Cards are generated from information stored as "notes". Notes are analogous to database entries and can have an arbitrary number of fields. For example, with respect to learning a language, a note may have the following fields and example entries:

• Field 1: Expression in target language – "gâteau"
• Field 2: Pronunciation – [sound file with the word "gâteau" pronounced]
• Field 3: Expression in familiar language – "cake"

This example illustrates what some programs call a three-sided flashcard, but Anki's model is more general and allows any number of fields to be combined in various cards.

The user can design cards that test the information contained in each note. One card may have a question (expression) and an answer (pronunciation, meaning).

By keeping the separate cards linked to the same fact, spelling mistakes can be adjusted against all cards at the same time, and Anki can ensure that related cards are not shown in too short a spacing.

A special note type allows the generation of cloze deletion cards. In Anki 1.2.x, those were ordinary cards with cloze markup added using a tool in the fact editor.

Anki supports synchronization with a free and proprietary online service called AnkiWeb.^[9] This allows users to keep decks synchronized across multiple computers and to study online or on a cell phone.

There also is a third-party open-source (AGPLv3) AnkiWeb alternative, called anki-sync-server,^[10] which users can run on their own local computers or servers.

Anki 2.1.57+ includes a built-in sync server. Advanced users who cannot or do not wish to use AnkiWeb can use this sync server instead of AnkiWeb.^[11]

Anki can automatically fill in the reading of Japanese and Chinese text. Since version 0.9.9.8.2, these features are in separate plug-ins.

More than 750 add-ons for Anki are available,^[12] often written by third-party developers.^[13] They provide support for speech synthesis, enhanced user statistics, image occlusion, incremental reading, more efficient editing and creation of cards through batch editing, modifying the GUI, simplifying import of flashcards from other digital sources, adding an element of gamification,^[14] etc.

While Anki's user manual encourages the creation of one's own decks for most material, there is still a large and active database of shared decks that users can download and use.^[15] Available decks range from foreign-language decks (often constructed with frequency tables) to geography, physics, biology, chemistry and more. Various medical science decks, often made by multiple users in collaboration, are also available.^[16]

Anki's current scheduling algorithm is derived from SM-2 (an older version of the SuperMemo algorithm), though the algorithm has been significantly changed from SM-2 and is also far more configurable.^[7] One of the most apparent differences is that while SuperMemo provides users a 6-point grading system (0 through 5, inclusive), Anki only provides at most 4 grades (again, hard, good, and easy). Anki also has significantly changed how review intervals grow and shrink (making many of these aspects of the scheduler configurable through deck options), though the core algorithm is still based on SM-2's concept of ease factors as the primary mechanism of evolving card review intervals.

Anki was originally based on the SM-5 algorithm, but the implementation was found to have seemingly incorrect behaviour (harder cards would have their intervals grow more quickly than easier cards in certain circumstances) leading the authors to switch Anki's algorithm to SM-2 (which was further evolved into the modern Anki algorithm).^[7] At the time, this led Elmes to claim that SM-5 and later algorithms were flawed^[17] which was strongly rebutted by Piotr Woźniak, the author of SuperMemo.^[18] Since then, Elmes has clarified^[7] that it is possible that the flaw was due to a bug in their implementation of SM-5 (the SuperMemo website does not describe SM-5 in complete detail), but added that due to licensing requirements, Anki will not use any newer versions of the SuperMemo algorithm. The prospect of community-funded licensing of newer SuperMemo algorithms is often discussed among users. However, there exists a greater focus on the development of the software itself and its features. The latest SuperMemo algorithm in 2019 is SM-18.^[19]

Some Anki users who have experimented with the Anki algorithm and its settings have published configuration recommendations,^[20] made add-ons to modify Anki's algorithm,^[21] or developed their own separate software.

In 2023 (version 23.10) the Free Spaced Repetition Scheduler (FSRS), a new scheduling algorithm, was integrated into Anki as an optional feature.^[22]

FSRS is based on a variant of the DSR (Difficulty, Stability, Retrievability) model, which is used to predict memory states. The default FSRS parameters are based on almost 700 million reviews from 20 thousand users and are more accurate in comparison to the standard SM2 algorithm, according to benchmarks, leading to fewer necessary reviews for the same retention rate.^[23][24]

The following smartphone/tablet and Web clients are available as companions to the desktop version:^[25][26][27]

• AnkiMobile^[28] for iPhone, iPod Touch or iPad (paid)
• AnkiWeb^[29] (online server, free to use; includes add-on and deck hosting)
• AnkiDroid^[30] for Android (free of charge, under GPLv3; by Nicolas Raoul)

The flashcards and learning progress can be synchronized both ways with Anki using AnkiWeb. With AnkiDroid it is possible to have the flashcards read in several languages using text-to-speech (TTS). If a language does not exist in the Android TTS engine (e.g. Russian in the Android version Ice Cream Sandwich), a different TTS engine such as SVOX TTS Classic can be used. AnkiDroid has also been used for other educational purposes. It is used as instructional media in Islamic Religious Education in Indonesia.^[31]

Damien Elmes, the Australian programmer behind the app, originally created it for learning Japanese.^[32][33] The oldest mention of Anki that the developer Damien Elmes could find in 2011 was dated 5 October 2006, which was thus declared Anki's birthdate.^[34]

• Version 2.0 was released on 6 October 2012.
• Version 2.1 was released on 6 August 2018.
• Version 23.10 was released on 31 October 2023, that has changed versions numbering scheme to year.month.^[35]
• Version 23.12 was released on 24 December 2023.^[36]
• Version 24.04 was released 31 March 2024.^[37]
• Version 24.06 was released 6 June 2024.^[38]

While Anki may primarily be used for language learning or a classroom setting, many have reported other uses for Anki: scientist Michael Nielsen uses it to remember complex topics in a fast-moving field,^[39] while others are using it to remember memorable quotes, the faces of business partners or medical residents, or to remember business interviewing strategies.

In 2010, Roger Craig obtained the then-all-time record for single-day winnings on the quiz show Jeopardy!^[40] after using Anki to memorize a vast number of facts.^[41]

A study in 2015 at Washington University School of Medicine found that 31% of students who responded to a medical education survey reported using Anki as a study resource; the same study found a positive relationship between the number of unique Anki cards studied and USMLE Step 1 scores in a multivariate analysis.^[42] In the same year, another study showed that students had a one-point increase on their licensing exams for every 1,700 unique Anki flashcards they used.^[43]

Another study in 2024 found that Anki was commonly used among respondents, with 86.2% reported using it and 66.5% using it every day.^[44] AnKing, an Anki deck developed by students at the University of Utah School of Medicine aggregates information from multiple third-party resources and has become the primary method of USMLE Step1 and Step2 study for many students, having been downloaded over 300,000 times as of 2024.^[45]

Anki offers user-made decks, which are commonly used in medical education and for learning a range of subjects including Chemistry, Biology, Geography, History, Law, Mathematics, Music, and Physics. User-made decks are also available for learning languages such as Arabic, Chinese, English, French, German, Hebrew, Japanese, Korean, Russian, and Spanish.^[46]

• Memrise
• List of flashcard software
• Computer-assisted language learning

• Gary Wolf (21 April 2008). "Want to Remember Everything You'll Ever Learn? Surrender to This Algorithm". Wired. ISSN 1059-1028. Retrieved 20 May 2022.
• Konrad M Lawson (28 April 2008). "Anki Review". Fool’s Flashcard Review. Archived from the original on 4 April 2009. Retrieved 23 March 2009. (part 2)
• Kevin Purdy (12 January 2009). "Anki Teaches Text, Audio, or Images Through Repetition". Lifehacker: Featured Download. Archived from the original on 2 April 2009. Retrieved 23 March 2009.
• Kristian Peltonen (24 September 2009). "New computer software makes studying easier by carefully timing reviews". Article on Anki & Smart.fm for the Canadian Press.^{[permanent dead link‍]}
• "Review of Mnemosyne vs. Anki vs. SuperMemo". Nihongo Pera Pera (Fluent Japanese). 1 June 2008. Archived from the original on 7 September 2008. Retrieved 8 September 2008.
• Glowing Face Man (22 February 2009). "Anki vs. Mnemosyne". Archived from the original on 26 February 2009. Retrieved 23 March 2009.
• David Harding (2009). "Mnemosyne and Anki". Ubuntu User magazine article.
• Baker, Stephen (2011). Final Jeopardy: Man vs. Machine and the Quest to Know Everything. Houghton Mifflin Harcourt. p. 214. ISBN 978-0-547-48316-0.

Wikimedia Commons has media related to Anki.

• Official website
• AnkiMobile Flashcards on the App Store
• SM2 Algorithm
• Anki Algorithm
• AnkiDroid Android package at the F-Droid repository