Every question has one intended answer and an explanation. Correct, incorrect, skipped and not-reached items are reported separately. The displayed denominator includes answered and skipped questions, not unseen items.
The HSK reference selector changes vocabulary annotations only. Sentence difficulty is not inferred from the highest-level word. The test has not been empirically calibrated to HSK, CEFR or school placement levels.
Questions were drafted with AI assistance and checked for answer consistency and script conversion. No independent native-speaker review is claimed. Practice recommendations use the skills associated with missed items.
The question set and vocabulary data are versioned. Repeated exposure and translated instructions can affect results; scores across languages or versions have not been equated.
The 15-question screen includes 12 vocabulary items (two from each of HSK 1–6), two grammar items and one reading item. The optional follow-up adds four different words per level. The vocabulary recommendation requires 80% correct at every preceding level; mixed results are flagged. This is a study recommendation, not a validated HSK placement. Listening, speaking, writing and HSK 7–9 are not assessed.
Each start samples a new form with the same coverage requirements and shuffles answer positions. Items seen in this browser tab are deprioritized. Grammar and reading practice sample six questions from fixed concept groups; vocabulary practice samples two words per HSK level. Coverage is controlled; equivalent difficulty across forms and languages has not yet been established with learner data.
Start the test