Under HSK 3.0 (launching July 2026), the speaking section is mandatory from Level 3. The previous HSKK was a separate optional oral exam; HSK 3.0 folds it into the core test. This guide covers the format, the rubric, and how to practice without a human partner.
Yes. Under HSK 3.0, the speaking section is mandatory from Level 3 onward. Levels 1 and 2 remain non-speaking; from Level 3 up, you cannot pass without recording spoken responses. This is the single biggest change for English-speaking learners moving into intermediate Chinese.
The Level 3 speaking section runs about 7 minutes total, broken into three task types:
Scoring is a 4-dimension rubric, each weighted equally (25%):
| Dimension | What's measured |
|---|---|
| Pronunciation and tones | Tones produced clearly; initials and finals distinguishable; no major segmental errors. |
| Fluency and rhythm | Speech flows without long unfilled pauses; rhythm is natural rather than choppy. |
| Grammatical accuracy | Sentences are well-formed; basic grammar (aspect, measure words, sentence-final particles) is used appropriately. |
| Vocabulary range and appropriateness | Word choice fits the topic; vocabulary scope matches HSK 3 expectations (~600 words). |
Each task is rated 0-5 by trained raters; the section score is the weighted average across all three task types.
Hanban has not published an absolute passing threshold for the speaking section in isolation. Official guidance: candidates should produce intelligible speech with adequate vocabulary and basic grammatical accuracy. In practice, scoring 60% or above on speaking is considered competent at Level 3, and 70% places you comfortably above the bar.
Speaking practice requires three things in tight succession: prompt exposure, immediate feedback, and a recording-replay loop. An AI tutor that listens, scores pronunciation, and replays your audio against a reference is the closest substitute for a human teacher. Aim for 15 minutes daily, 8-12 weeks before the exam — consistency beats long sessions.
Hanlexon's Prof Hanlexon voice mode runs the rubric live: it listens, scores tones and fluency, and offers correction in real time. The HSK 3.0 mock exam includes timed speaking sections so you can practice the full format under pressure.