日本語

If you used Dropbox with previous versions of Anki, please read the next section before upgrading.

If you ever exported & imported .anki decks in the previous version, or find you are missing cards after upgrading, please read this section.

Reviews are limited to 100/day by default.

When you answer "again", there are now two learning reviews of the card, so the learning (orange) count will increase by two.

Cloze deletion has been much improved, but now requires you to select a special cloze type (the top left button in the Add window).

Selective study is still possible, but is done differently. You can either move cards into different decks (via Change Deck in the card browser), or create a filtered deck that searches for particular tags, card properties and so on.

If you study material in multiple directions / used multiple card templates, please see this section for info on how cards are selected now.

Anki only checks the first field to see if it’s unique now.

If you delete a deck and sync, that deck will now be deleted on your other devices as well.

If you use "hide" on any decks in Anki 1, Anki 2 will not automatically import them. You can still import them manually, via File>Import. Anki 1 decks were stored in Documents\Anki by default.

Instead of storing each deck in a separate file, Anki now stores all of your decks in a single file called a collection. This makes it easy to move cards between decks, and different decks can now share models and facts.

Decks can now be organized in a tree. Use "::" to mark descending levels, like "German::My textbook::Lesson 1".

Multiple decks can share the same settings group, so updating the settings for one deck updates the settings for the other decks using that group.

If more than one person uses Anki on your computer, you can now create a separate user profile for each user. Each user profile has their own collection of decks, and Anki will remember settings separately for each user. More info.

A fact is now called a note.

Instead of talking about a note’s model, we now talk about a note’s type.

A collection stores all your notes, cards, and so on.

You can divide the cards up into separate groups, called decks.

Plugins are now called add-ons.

The new cards per day limit is controllable per deck, and bounded by the parent decks. This means if "French" has a limit of 20 cards and "French::Lesson 1" and "French::Lesson 2" both have limits of 15 cards, you’ll get 15 cards from lesson 1 but only 5 cards from lesson 2.

Decks now have a limit on daily reviews as well, which defaults to 100. When the limit is reached, Anki will not present further reviews from that deck, even if there are cards waiting. When the congratulations page is reached, a notice will be displayed, advising the user that for optimum memory, they should consider increasing the daily limit. This change is intended to reduce the sense of being overwhelmed that people feel after returning to Anki from a break. Large due counts can be off-putting, and while it’s not optimal for a user to leave due cards waiting, if they give up in frustration that’s an even worse outcome.

No limit is applied to cards in learning, as the learning count is a product of the number of new cards and reviews that are done each day.

Shared decks can now be browsed on AnkiWeb, and downloaded from your browser.

You can now rate and comment on shared decks.

You can now pull in new cards from a previously downloaded shared deck.

Nested categories like "French→Book 1→Lesson 3" are more intuitive to the typical user than the tags "french book1 lesson3".

Many users struggled with the old selective study system, be it because they’d accidentally used overlapping tags that cancelled out what they wanted to do, or because they’d removed particular tags from review and forgotten they’d done so.

Tag-based limiting did not perform well - slowdowns of 10x or more were not uncommon.

Some users wished to assign different study settings to different cards. This is not practical with tags, as they overlap, and it would require an elaborate system of defining which tags should take precedence for that to work.

In Anki 2, the standard review system is based on the current deck, rather than a list of allowed/disallowed tags. Decks can be nested, allowing you to define the order cards should appear in and allowing you to focus on specific subsections individually.

Tag-based study is still possible via a flexible filtered deck system, described further below.

Like selective study, card display order is now controlled via nested decks instead of tags. When you select a given deck, Anki walks the deck and its subdecks in alphabetical order looking for cards. So if you have cards in "French::Vocab", "French::Sentences" and "French:Misc" and you select the "French" deck, cards will be fetched first from Misc, then Sentences, and then Vocab, as that is the alphabetical order of those decks. You can use this property to ensure certain cards are reviewed earlier or later than normal: prepending a "!" like "!Vocab" would ensure vocab cards come first, and prepending a "_" would ensure they come last.

For more fine-grained reordering of new card presentation order, the browser now has a reposition command which can change the order of new cards in the queue. You can use this for prioritizing material, changing the order cards of different types will be mixed together, and so on.

Anki has a much-improved learning mode. When new cards are introduced or old cards forgotten, they can now be shown over a number of increasing intervals before they graduate into normal reviews cards. More information in the manual: learning

Reviews are only scheduled on a per-day basis now. If you wish to divide a day’s cards into multiple sessions, you can either use the timeboxing feature or simply do a certain number of cards and stop until later.

Only when you fail a review card is the card’s lapse count increased. This means that failures while the card is in learning no longer count towards a card being a leech.

You can customize the behaviour when failing reviews. By default failed reviews are put back in the learn queue to be relearnt. You can have a different set of steps for cards that are being relearnt, which allows you to reduce the steps required or require longer waits compared to cards that are being learnt for the first time.

A multiplier is now available to cause Anki to schedule reviews more frequently or less frequently than it would otherwise.

Due counts are now displayed in the more logical new → learning → reviewing order. The ETA and progress bars have been removed, and the timer is off by default (in the above screenshot, the due counts are not shown as they are only shown while the question is visible)

The learning count is the count of how many answers are required to complete the queue, not how many cards. With the default settings of two reviews to graduate, the learning count will be twice the number of cards to learn.

In normal study, reviews are always sorted by due order now. If you need to change the order to deal with a backlog, you can do it with a filtered deck.

While there is no explicit option for "show failed cards at end" anymore, you can achieve the same behaviour by setting the learning steps to a length longer than your session length (eg, 1 hour), and increasing "learn ahead limit" in the preferences to that number as well.

Filtered decks allow you to gather cards from other decks based on specific criteria and with a specific ordering.

They are very flexible: they can be used for previewing cards, cramming them before a test, studying particular tags like the old selective study feature, catching up on a backlog with a particular sort order, reviewing ahead, going over the day’s failed cards, and more.

They can be set to reschedule based on your answers (like Anki 1.2) or leave the original cards alone (like Anki 1.0).

Filtered decks also replace the old "review early" button. They offer two advantages: you can now choose how many days to look ahead, and you can close Anki and resume reviewing early where you left off, instead of starting with the same cards again like in Anki 1.2.

For more information, please see http://ankisrs.net/docs/manual.html#filtered

In most cases, you can now merge changes from multiple devices instead of having to choose which side to keep.

Full syncs are only necessary on the first sync and after changing things like the number of fields in a note. If you’re about to make a change that would force a full sync, you’re warned about it and given the option of aborting.

Modifying many cards or notes no longer forces a full sync - incremental syncs can now handle even large amounts of data.

Clock drift will no longer lead to incorrect syncs. That said, it’s still confusing to move to another machine with a different clock and find greater or fewer cards due, so Anki will still require the clock to be within 5 minutes of the real time.

All data is encrypted before being sent over the internet.

If the connection times out, Anki will automatically retry.

AnkiWeb now supports media natively, instead of requiring a third-party service like DropBox. No special setup is required - when you sync your decks, their media will be automatically synced with it. Anki monitors the media folder now, so there are no issues with an out of date media DB like there were in the past.

Media and LaTeX references in templates are no longer supported, because they don’t play well with shared decks. The upgrade routines will automatically create a new field with the media tags embedded in it, like [sound:apple.mp3]. If you manually add the references back in, Anki will not notice the media is in use when you check the media DB. Please note that if you had media references comprised of more than one field (eg automatically and you will manually have to sort this out.

If you had static media references like <img src=logo.jpg> in your templates, those media files will not be imported and you’ll need to manually copy them over. It’s recommended that you rename them to have a leading underscore, as doing that tells Anki they are used in the template and shouldn’t be deleted when checking for unused media.

When you upgrade, Anki moves the media in each deck’s .media folder into the new collection’s media folder. You had different files with the same name in multiple decks, Anki will automatically rename the media for you.

If for some reason Anki wasn’t able to move your media from its old location, please manually move the media files from the old .media folders into your Documents/Anki/User 1/collection.media folder.

Tags and decks are displayed in a panel on the left, and can be clicked to search for. Hold down ctrl or command while clicking to append the item to the search string.

Columns are now configurable - right click on the header and you can enable and disable the columns you want to search on. Click on a column to sort by it; drag a column to move it to a different place.

Previously you could ask the browser to sort by a particular field, but doing so caused any notes without that field not to show. In 2.0, each note type has a "sort field". The sort field can be shown as a column, and can be sorted on. Click the "Fields…" button to change the sort field. Anki will automatically sort in numeric order if the field contains only numbers; you don’t have to tell it a field is numeric anymore.

Many search operations are faster than before, but some cases are slower. The search box now requires you to hit enter; it will not automatically search as you type, as some people type slower than others and for expensive searches, automatic searching can be painful.

In 1.2, a copy of each field was kept with style information stripped, so that you could search for a word and have matches appear even if part of the word contained style information (bold, etc). Anki 2 drops this cache, as it was expensive to maintain. In Anki 2 if you want to search for partially-styled text, you need to limit yourself to the current sort field (which Anki can use when searching)

Selections are now preserved when the search criteria is changed, so you can do things like select a bunch of cards, then click on "New" in the side bar to narrow the selection to only cards that are new. Preserving the selection of many cards at once is slow however, so if the selection is too large it is discarded on the next search.

Rescheduling behaves differently now. Cards have their interval and ease reset but keep their review history. If you need to remove a card’s review history, please search for rescheduling in the manual.

The browser now exposes the full review history. Click the info button on the toolbar to see it. Note that this data was stored differently in previous versions, so the display of whether a card was being learnt, reviewed, relearnt or so on may not always be correct for old data.

In Anki 1.2, {{ field }} references were automatically wrapped in style information. This meant that the prompt text and field text of a template like "What is the capital city of {{ country }}?" often ended up different, and users had to use {{{ field }}} instead and figure out the necessary HTML.

In 2.0, there is no implicit styling: instead of choosing fonts, colors, alignment and so on in the GUI, it’s all embedded in the template as standard HTML and CSS. As a bonus, this allows you to do things like use a gradient or image for your card background if you wish.

The "hide question when showing answer" option has gone away: Anki always replaces the question with the answer now. Templates default to including the question fields on the answer text however, so Anki will behave like always. The "show divider between question and answer" option has gone away too, as the divider is included on the template and can be removed if desired.

In previous versions of Anki, there was a button in the top right of the add cards screen that let you enable and disable card templates. As the button would only control which cards would be added when you added new content, if you wanted to add extra cards to previously added content, you had to use the "generate cards" feature in the browser. This caused a lot of confusion.

In Anki 2, instead of manually specifying which cards to create, cards are created depending on whether fields are empty or not. By default, a card will be created if at least one field on the front template is non empty. You can set Anki up to only create particular cards if certain fields are not blank.

If you wish to selectively add cards (such as adding a reverse card for only your difficult content), you can add a field to your notes that controls whether the reverse card will be generated. If you put any text in the field (such as "y"), the card is generated, and if you leave the field blank, the card is not created.

When Anki 2 upgrades your old decks, it will automatically add such a required field to notes where you have a card enabled for some but not all of the notes.

There is no option to manually delete individual cards in Anki 2, as if you manually deleted a card, Anki would automatically add it again if fields were not empty. Instead, you can remove individual cards by blanking out one of their required fields.

Anki doesn’t immediately remove cards when one of their required fields is blanked out, because that could lead to data loss. To remove cards with missing required fields, run Tools>Maintenance>Empty Cards.

As blanking out fields is a few more steps than just pressing the delete button, if you need to remove a number of cards during review, you may find it easier to suspend them instead, and then use find&replace at a later date to blank out a required field for all the suspended cards in one go.

For more information about selectively adding cards, please see this section of the manual.

You can no longer specify whether a field should be unique or required. Anki makes sure the first field is both provided and unique, but does not check the other fields. Anki did not perform well when many cards were marked unique (since it has to cache all that information), and a single field is sufficient for determining if a given subject has been entered in before. If you wish to look for accidental duplicates in secondary fields due to typos, you can use the Find Duplicates feature in the browser.

For more info, please see this question.

Fields can now be marked sticky. This means that when you add notes, the field content is not erased and is left in the field. This is useful when you’re adding a series of notes that have similar field content.

Decks are considerably smaller now. Most people should see a size drop of 3-6x.

Many operations have been optimized. You should find your decks load much faster now, and studying should be smoother. These changes should be especially pronounced on the mobile clients.

Anki should now perform much better when your hard disk is being heavily accessed.

Graphs load much faster, and it’s now possible to see graphs of your entire collection at once.

Fields now dynamically expand to fit their content.

Fields can now display complex HTML without mangling it.

You can now jump to a conflicting note when conflicts are detected

Images are shown at a reduced size in editing mode.

Cloze deletion now makes it easy to generate multiple clozes from a single body of text. All of the clozes share the same note, and edits to the note will update all cards at once.

Cloze deletion works with LaTeX now, too.

Clozes appear as bold and blue like before, but they can now be customized in the template.

Clozes made with previous versions will not be converted to the new style; it only applies to newly made cards.

The deck statistics and graphs have been merged into a single report that contains both text and graphs, and they have been completely rewritten in the process. There’s now more information available, and it’s easier to compare things like the cumulative work spent on reviews vs relearning vs initial learning.

The old period selection boxes have been replaced with a single box down the bottom, allowing you to change between a monthly, yearly, and deck life view of the data. The data is chunked into weeks or months in the larger graphs, which both performs better and offers a clearer view of trends.

There is a button to save the graphs to an image, to make it easier to share your graphs with friends or teachers.

When you start studying, the new cards for the day are gathered - 20 by default. When each card is shown, all its siblings are moved to the end of the queue, so they’re spaced as far apart in those 20 cards as possible.

If you want new siblings to be introduced further apart, you can put them in different decks or use the reposition feature of the browser.

When determining the next review time of a review card, Anki now checks the surrounding days and tries to place the review on a day when no siblings have reviews. The possibilities to consider are configurable.

The "save" and "save as" operations have been removed. Saving is now transparent to the user, and happens every 5 minutes of adding/reviews, and after various other operations. By having more control over the saving process, Anki can save more efficiently, and revert changes when an operation goes wrong.

The undo system has been simplified. The old method was comprehensive, but made updates to the deck considerably more expensive, and could cause inconsistencies under rare circumstances. Multi-level undo has been preserved for reviews, but other destructive operations now have a single level undo, which is guaranteed to restore the deck to a consistent state.

In 2.0, Anki will be moving from the GNU GPL3 license to another GNU license: the AGPL3. The Affero license is compatible with the GPL3, so existing plugins do not need to be relicensed. The major difference from the GPL3 is that the license requires people who modify the code and run it on a web service to make that modified code available to users of the web service. It effectively means a company can’t make an online service using Anki’s code unless they reciprocate by sharing their changes. This clause should not affect regular users at all.

Also of note is that ankiqt is now dual licensed like libanki was. I don’t have any immediate plans to take advantage of this, but it opens up the possibility of custom versions in the future for companies who want to distribute a stripped down and/or content-protected version of Anki with their content for a fee. The desktop version will continue to be released as open source, and I have no plans to change that.

Because much of the code has been rewritten in 2.0 there is very little code contributed by other people remaining. I’ve contacted the authors of the remaining code and asked for their permission for the licensing change. I don’t think I’ve missed anyone, but if I have, please let me know.

Because of the extensive changes, almost all add-ons will require some level of modification to work with 2.0. Some add-ons have been ported already, others will likely come in the future.

After upgrading, please visit the add-ons site to look for new versions of the add-ons you were using in Anki 1. The add-ons site is available from the Tools>Add-Ons menu inside Anki 2.

There’s now an add-on writing guide here that also covers the main things to be aware of when porting an old add-on.

Download this zip file

Open Anki 1.2, and go to Settings>Plugins>Open Plugin Folder

Extract the .py file from the .zip you downloaded, and place it in the plugins folder.

Restart Anki 1.2

For each deck that didn’t import, open it and choose Tools>Make Cards Unique.

The SuperMemo algorithm was not designed to work on a sub-day basis, and is not accurate to that level.

In previous versions of Anki, in order to review in the evening and then again the next morning, cards had to be scheduled with small intervals like 8 hours. Unfortunately this meant that cards reviewed in the morning would become due again in the evening, when they are better reviewed after a night’s sleep.

By making all of a day’s reviews due in the morning, you can both see at a glance how many cards are due that day, and study as many or as few as you desire. When you have a busy morning you can leave the cards until later, and likewise when you have a lot of free time, you can do all the cards in one session. When cards become due throughout the day, it’s not easy to find out how many more need to be done that day, and you have to jump through hoops to study more than are currently due.

For the initial learning steps where sub-day scheduling is more important, you can use Anki’s learning mode instead.