This is a third-party app and is not created by the WaniKani team. By using this, you understand that it can stop working at any time or be discontinued indefinitely.
I always used wkstats but wanted something that worked better on my phone. So I built Wanilog: a phone-first WaniKani stats dashboard that runs in your browser, installs like an app, and keeps working offline after the first sync.
You can try it without an API key: hit “Try the demo” on the homepage.
Features
Finish-date projection
Forecasts when you’ll hit level 60, or any goal level you set, from your own per-level history. You get a median pace, a confidence range, and milestone dates you can export to your calendar.
How much Japanese you can read, kanji by kanji, weighted by frequency across the top 2,500. Paste any text into “Can I read this?”, or skim live NHK headlines colour-coded by what you already know.
Your Guru’d kanji counted against three lists: the JLPT levels, the 2,136 official Joyo kanji (broken down by school grade), and the 2,500 most common kanji.
Your correct rate per item type and per SRS stage, plus an effective pass rate (\text{meaning} \times \text{reading}) and an estimate of the hours you’ve put in.
Every item you’ve studied, grouped by SRS stage from Apprentice through Burned. Filter by part of speech when you only want to look at verbs, or nouns.
There’s also an optional public profile page, a ⌘K command palette, a full Japanese UI, and four colour themes.
Your API key stays in your browser and your synced data lives in your own IndexedDB. A few opt-in features send the key to a Wanilog server. It’s free, with no ads and no account.
Install it like an app
It’s a real PWA. Settings → Install app walks you through it for whichever browser you’re on, on iOS, Android or desktop. Once it’s installed it launches from its own icon and runs offline.
I’ll try it for a while, but right away it seems nice. First thing I thing I’d like to see changed would be the “What if you were faster?” …. It has me averaging 21 days per level. Lately I’ve been slower than that. It would be nice to see what 25 days per level looks like. I’m definitely more likely to ponder going slower than faster at this point.
I like it so far, just wish the burn vs not learned on the kanji map were easier to tell apart. Since both are grey, it’s a little confusing. I liked the yellow that wkstats uses.
I discovered the “Reading coverage” page which is pretty interesting. For NHK Easy news it shows a bunch of 100% coverage headlines that are all quite old. I’d love a way to see the best articles (for me) over the last n weeks.
Thank you! I also noticed when I was looking at the kanji map for jlpt, that the last couple hundred n1 kanji there is no more referencing to Wanikani. Are those kanji not on Wanikani? Would it be possible to put those separated like wkstats does? It’s a little abrupt for the kanji to suddenly not say anything or link to anywhere.
No worries. Unfortunately the NHK Easy website has no reliable way to get the latest articles, so I decided to just remove this for now. I’ve added some new options to the regular NHK - you can now filter by “Best coverage / Best for learning / Most recent”. Hope that helps
Hey everyone, I have added a new sharing feature - you can now make a copyable “Spotify Wrapped” esque card which you can share with others. Example below:
I disagree. Accuracy in our srs reviews simply reflect what % of questions were answered correctly. So, if we were tested on 10 meanings and 10 readings and we got 9 meanings correct and 7 readings correct, we get 90% for meanings and 70% for readings. And, in total we got 16 items correct out of 20 items for an accuracy of 80%.
Multiplying the 90% and 70% figures to get “total” or “effective” accuracy occurs in the realm of probabilities where the outcome of the subsequent event is dependent on the outcome of the first event which isn’t the case in our srs reviews.
Please return your accuracy calculations as you had it before.
My interpretation is different, since I consider one kanji (or one vocabulary item) as a unity of meaning AND reading(s). So in your example: If we tested 10 meanings and 10 readings it is still only 10 items that got tested.
Or in other terms: Only when I can recall the meaning AND reading of an item, I consider it as learned. It is indeed similar like probabilities: Accuracy is the probability that I can remember the meaning AND reading of a kanji/vocab.
This is a third-party app and is not created by the WaniKani team. By using this, you understand that it can stop working at any time or be discontinued indefinitely.
Wanilog, a WaniKani stats dashboard (wanilog.com)
