Wanilog vs wkstats
wkstats, built and maintained by rfindley for the better part of a decade, has been the WaniKani community's stats home since the mid-2010s - every dashboard in this space, Wanilog included, is downstream of it. Wanilog keeps the projection, JLPT, Joyo, and frequency math you already rely on, and adds a confidence band on your finish date, reading coverage on real Japanese text, goal tracking, and a phone-first PWA that works offline.
What wkstats got right
- It defined the category. Finish-date projection, JLPT coverage, Joyo grade buckets, frequency lists, time-on-level history. Every WaniKani stats tool, Wanilog included, is downstream of that work.
- Per-level item drill-downs. Still the canonical reference for browsing radicals, kanji, and vocab one level at a time.
- It got the fundamentals right. The projection and coverage foundations were solid enough that a decade on they are still the model the rest of us, Wanilog included, measure ourselves against.
Side-by-side feature matrix
| Feature | Wanilog | wkstats |
|---|---|---|
| Setup and hosting | ||
| Hosted, no install | Yes | Yes |
| Self-host option | No | No |
| Open source | No | Yes |
| Demo mode (no API key) | Yes | No |
| Stats and analytics | ||
| SRS stage distribution | Yes | Yes |
| Review and lesson forecast | Yes | No |
| Workload forecast (lessons-per-day slider) | Yes | No |
| Peak review day surfaced | Yes | No |
| Per-level accuracy chart | Yes | No |
| Leech list with severity | Yes | No |
| Leech trainer (multiple algorithms) | Yes | No |
| Confusion pairs / visually similar items | Yes | No |
| Burn velocity (week / month) | Yes | No |
| Critical-kanji-to-level-up panel | Yes | Partial |
| Coverage and forecasting | ||
| Level 60 finish-date projection | Yes | Yes |
| Statistical projection with confidence interval | Yes | No |
| Goal level (1-60) and goal date | Yes | No |
| JLPT N5-N1 coverage with drill-down | Yes | Yes |
| Joyo coverage (grades 1-9) | Yes | Yes |
| Frequency-list coverage | Yes | Yes |
| Reading coverage on live NHK headlines | Yes | No |
| "Can I read this?" arbitrary text tool | Yes | No |
| Manual known-kanji input | Yes | No |
| UX and device support | ||
| Installable PWA | Yes | No |
| Offline shell | Yes | No |
| Phone-first layout | Yes | No |
| Dark mode | Yes | No |
Last verified: July 2026. Competing tools evolve, so a claim may lag behind a recent release. Corrections are welcome.
wkstats' clearest edge in the matrix is published source you can run yourself; most of Wanilog's extra rows are forecasting depth and phone-first packaging.
How Wanilog builds on it
The finish date becomes a range
wkstats averages your past level durations, which is a solid baseline. Wanilog adds a p25-p75 confidence band on top, so a single slow week (a holiday, an illness) never shifts your finish date by a month. You see a range you can plan around instead of a single brittle number.
Coverage checked against the news
wkstats scores your kanji against frequency lists. Wanilog takes the same idea to live text: NHK Easy and NHK News headlines pulled fresh, each kanji you have taken to Guru highlighted, and a readability percentage shown alongside. Paste any Japanese text into Can I read this? and get the same score back on the spot.
A dashboard that opens on the train
Wanilog installs to your home screen on iOS, Android, and desktop. The Service Worker keeps your last-synced data available without a connection, so the dashboard opens just as fast on the train as it does at your desk. Every page was laid out for a phone first.
A goal with a deadline
Pick a target level (1-60) and a target date. Wanilog shows an on-track or behind banner across the home, projection, and coverage pages, so you know where you stand without doing the math in your head every week.
Leech trainer and workload forecast
Two extra surfaces that go beyond the core charts. The leech trainer runs three scoring algorithms so you can pick the one that flags the items you actually struggle with. The workload forecast graphs the next 60 days of expected reviews, so a heavy week becomes something you see coming rather than discover the morning of.
When wkstats might still be the right pick
- The layout is in your muscle memory. Years of looking at the same dashboard is real value; if you only check projection and JLPT once a week, you may not need anything more.
- You want the original. wkstats predates Wanilog by roughly a decade and carries a decade of community trust.
- You want the source code. rfindley published the wkstats source in early 2026, so you can run your own copy or tinker with it. Wanilog is not open source.
FAQ
Is Wanilog a replacement for wkstats?
It can be, though it does not have to be. Wanilog covers the same core ground - finish-date projection plus JLPT, Joyo, and frequency coverage - and is actively developed, so it works as a full replacement if you want one. wkstats still runs fine, and plenty of people use both.
Will my wkstats projection match Wanilog’s?
Not exactly. wkstats averages your past level durations; Wanilog uses the median plus a p25-p75 confidence band. On steady users the numbers stay close; on users with one or two slow levels Wanilog is more stable.
Can I see a per-hour review heatmap?
No, and neither can wkstats for new users. WaniKani deprecated the per-review endpoint a few years ago and the API now returns an empty array there. Both tools fall back to per-subject review statistics, which give accuracy and counts but not timestamps.
If you want to see your own numbers in it, open the demo for a feel of the layout, or sign in with your API key - a full history syncs in under a minute.