Sorry I don’t have anything specific to hand, but next time I take a test I’d love to give feedback.
How/Where is the best place to reach you?
P.s. and Bunpro is the only place so far where I’ve found any feedback after I’ve taken a test, so it’s not a criticism, I just want more!
Feel free to contact me directly on the forums or to send feedback through any of the feedback reporting forms on most pages of the site.
Wow, I just checked this webpage, is it legit? So many practice tests! 1000 n3 grammar questions!!!
It reminds me of my high school years, when I solved entire math practice books, becuase I was so bored and intellectually starved, like 300 pages of math practice questions 
Challenge accepted.
I suppose it is the grammar questions from different jlpt exams gathered in one doc, right?
Thank you, I appreciate the quick feedback.
I think it’s still completely broken, or my profile is somehow bugged. I have done several of the lower level tests, and all the N1 tests, and one was rather recently, but I still don’t have any jlpt tag and there isn’t an option to toggle it in profile settings either.
Not really an issue since that’s not the point of the tests, but these kind of bugs always get stuck in my mind. It might help to run a cron job every other week or so to go through past tests and award anyone who had connection issues at some point. Then it doesn’t really matter if it’s reliable or not.
You’re right. It wasn’t just a connection problem. There’s a bug too.
I think I found the cause and it should be fixed in our next update.
Anyway, we updated your profile manually so it’s displaying the proper level already.
Thanks for pointing this out.
I also have the issue 
No need to update it manually, but I’ll be sure to check it again soon if it is to be fixed in the next update 
Hi Asher, I’m following up on our conversation about feedback for the Bunpro JLPT answer explanations. I’ve just taken the 5th JLPT N5 test, and there was an answer/explanation that wasn’t clear to me.
First section (Vocab), question 17.
I struggled with translating the sentence in the test. Some of of the vocab and phrases in this sentence aren’t covered until N4, and I hadn’t come across them. Nodo = throat, and the phrase “nodo ga kawa ku = to be thirsty” are N4 vocab and phrases.
My thought process during the pressure of the test situation: kawaitakara, this might be some grammar point to do with “Kawaii” and “ta” and “kara”? I also misread nomimono as “no” and “mimono”. I thought “mimono” might be “something to see”. In the end I went for “yomitai” because my best guess for the answer was thinking that I “want to read” something that I see.
Obviously I was wrong for several reasons, but here is where I feel Bunpro could improve this section:
Even looking at the mark scheme, it wasn’t clear from my existing knowledge why “nomitai” was the correct answer. The explanations for why each possible answer was right or wrong don’t clarify anything if you don’t know what the sentence says, and there is no way on Bunpro of finding this out.
I ended up copying the sentence and putting it into google translate and asking the AI to parse the grammar.
A translation of the sentence itself in the answer section would be helpful, and possibly a link to the grammar point “Nodo ga kawa ku”, especially because (a) it’s a set phrase and (b) it isn’t covered in Bunpro’s N5 content.
I’ve only gone through the answers to the questions I got wrong, so it’s not a critique of the entire test, just about 5 of the answers.
It’s tough trying to cover every base, and if you are setting the questions and know the right answers, it can be hard to see or predict where everyone will go wrong. Only 1 answer is correct, but people can be incorrect in a million different ways. Sometimes it’s only when you can see someone else’s thought process that the gaps become clearer. So I hope this is helpful!
You made a couple of interesting points here.
At the moment, the way the JLPT questions are set-up is to give the test taker advice when they have selected something wrong, and to tell them why that particular answer is wrong in context with the question. In many cases, as you highlighted, it is hard to account for every word in a sentence that a user may not know, and if that was the particular word that tripped them up.
After having a brief discussion with the team about this, we are quite possibly going to add English translations to most if not all of the test content, in order to remove at least one of those variables.
This should work for quite well for single-sentence questions, but I have a feeling it will still be quite difficult for entire reading passages where even having the text might not make the answer immediately obvious. In those cases, we will just need to continue doing our best to predict where the user actually went wrong, and highlight those points in the official hints.
Thanks for the in depth reply.
I agree Bunpro can’t (and shouldn’t have to) account for every infinite variety of potential mistakes users will make, but there just needs to be a little something extra to help users to find out for themselves where they went wrong.
English translations would be good. Or a romaji translation with word spaces. Just something so a user (using my own experience) can slap their forehead when they realise it was “nomimono” and not “no mimono”, or give them a word (“nodo”) they can then search for on Bunpro’s search function.
The search itself will help users to learn, and if it can all be accomplished through Bunpro then that’s even better.
One thing I want to say in defense of bunpro here, and I know this isn’t the main thrust of your post, is that it’s good they incorporate ‘non-N5’ material.
Despite there being a sort of agreed-upon canon of grammar points and vocab for each JLPT level, they do not release a list of items that will be included on each test or really offer much in the way of guidelines for what to study. The stuff included changes every year, and all we have are guesses. So you have to be prepared for things beyond the scope of what you expect.
Honestly, these tests have long been criticized for being too easy. N1 in particular is regarded as a very easy test for what it supposedly measures. I would not be overly shocked if one year they responded to those criticisms and raised the difficulty across the board without telling anyone first
I would like to request that you don’t add English translations to the n3 tests and above… I think from that point it becomes more of a crutch than a help. Part of doing practice tests is also doing your own problem solving afterwards, as well.
Yep, I feel like you can just use problem solving from day one and just decode Japanese like it’s some kind of ancient Egyptian
how about just having them hidden by default?
I still dont think they really add much value tbh, seems like a lot of extra work
also I think translation often leads people astray
you know, i thought i heard somewhere that this year’s JLPT tests were going to be “different” somehow. i guess we will know soon.
Translations would be hidden by default and only available at the same time one receives hints for the answers, so it wouldn’t be taking anything away from the experience from those that want the harder, more realistic experience anyway.
TBH I dont even want the temptation HAHA but fair enough
After the N4 exam last December, I gave a little bit of feedback in this thread. My conclusion was basically that the difficulty of Bunpro’s N4 mock tests was precisely right, though the Bunpro listening questions about times were harder than the one on the real exam.
If I recall correctly, there were many sentiments that the N3~N1 mock tests on Bunpro were too easy compared to the real exams. I also recall, however, that some were wondering if the 2025 December JLPT was an outlier in terms of difficulty. Some were wondering if it was a strangely hellish exam that doesn’t properly reflect average JLPT difficulty.
I just took the 2026 July N3 and thought I’d share my experience on how it compared to the Bunpro N3 mock tests. I reckon the more data/feedback there is, the better Bunpro can finetune.
My experience was:
Detailed descriptions:
For the vocab section, I don’t think any changes are needed. I believe the reason the real test was slightly more difficult is simply that the Bunpro tests mostly follow the Bunpro decks, which I studied extensively in preparation for the JLPT. The real test contained a handful of words I did not know, but I’m somewhat convinced this is just unavoidable. It’s probably impossible to accurately account for every single piece of vocab the official exams throw at you. Interestingly though, one of the words that was specifically tested for is a word that Bunpro has listed as A8. The word happens to be in my review pile.
For the grammar section, the star questions tested some grammar I was not familiar with, and one fill-in-the-gap question specifically tested for a grammar point that is in the Bunpro N2 deck (!). I happened to have only just studied it in the week before the exam. The long texts were about the same level as the Bunpro mock tests, though they might have used slightly harder vocab, I feel. No real discrepancy there. I feel the biggest differences were in the fill-in-the-gaps and star questions.
For the listening section, the discrepancy was very big. The real test had much faster speech and much more high-level grammar and vocab. The Bunpro mock tests are not reflective of this summer’s N3 test. However, talking with people in the JLPT July 2026 thread, the general concensus seems to be that the listening section was extremely hard, so maybe it was a bit of an outlier this year. No way for me to know.
All in all, the Bunpro mock tests were still a fantastic help in preparing for the JLPT. I sure would have been so much more ill-prepared if I hadn’t had Bunpro.
I took the same N3 test as you did too. I concur with everything you said.
For grammar sections, I feel that the texts were all almost twice as long as the ones in Bunpro.
For listening, I felt like they were using vocab I’ve never heard before although I’m at 10k words with my Anki.
The mock tests definitely helped me prepare and anticipate the structures of the real tests.