Chinese text analyzer
Paste Chinese text. Get its character count, historical HSK character coverage, and the characters outside your selected HSK set. Nothing is uploaded — the analysis runs in your browser.
Analyzer
Up to 50,000 UTF-16 units (most characters use one; emoji and rare Han may use two).
Replaces the passage after successful validation. UTF-8, or UTF-16 with a byte-order mark; up to 200,000 bytes and the passage limit above. Read only in this browser.
Personal character coverage
Up to 50,000 UTF-16 units. Each Chinese character counts as known, including characters inside pasted words. Your HSK level adds a baseline; with “Not set”, only this list counts. Simplified and traditional equivalents share coverage. This list and your passage are never saved or uploaded.
Chinese characters
0
Unique characters
0
HSK level
—
All characters
0
Character coverage by HSK level
Every Chinese character in your text, filed under the lowest HSK level that introduces it. Percentages are of Chinese characters, counting repeats.
| Level | Characters | Share | Cumulative |
|---|
Characters above HSK 6
Not in any HSK 1-6 list, most frequent first. Proper nouns and technical terms usually land here, which is why a high count is not automatically a hard text.
Includes every character in this list, frequency, and its historical HSK introduction level. Unknown levels are labeled outside HSK 1–6. Original character forms are preserved. This is a character list; definitions and pronunciation are not included.
HSK vocabulary matched
Personal character coverage
Highlighted characters are outside your baseline and known list. This measures character familiarity, not word knowledge or comprehension.
All unknown characters, most frequent first, with original forms and HSK levels. No passage text or known-character list is included.
Reading time
Export this analysis
A CSV of everything above — the summary, all six level rows, the characters outside the selected HSK set, and every HSK word matched — with the method, the word-list version and a SHA-256 of the text you analysed in the header, so a reader can tell that two results describe the same passage. The file is built in your browser; the text still goes nowhere.
Free to reuse under CC BY 4.0 with attribution to this page.
Chinese character counter: what the numbers include
The Chinese character count includes Han characters in both simplified and traditional text. Punctuation, spaces, Latin letters, and digits are excluded from that count. The total text length includes all Unicode code points, while the unique count includes each distinct Han character as pasted only once.
For example, 你好，你好！ contains four Chinese characters, two unique
Chinese characters, and six code points in total. This is a character counter, not a
segmented Chinese word counter. The HSK vocabulary matches below the results are a
separate, incomplete list of recognized words.
How the level is decided
The tool reports the lowest HSK level at which at least 90% of the Chinese characters in your text are covered by the HSK 1-6 character sets. If no level reaches 90%, it says the text sits above HSK 6.
Two choices in that sentence are worth stating plainly, because they are what makes the answer reproducible rather than a black box:
- Characters, not words. Word segmentation in Chinese is itself an estimate, and two segmenters disagree on real text. Characters are countable without judgement, so the number does not depend on which segmenter this page happens to use.
- 90%, counting repeats. Coverage is measured over every character occurrence, not over the distinct set. Repeating familiar characters raises this percentage even when an unfamiliar character remains. It measures character occurrences, not comprehension.
Each character is filed under the lowest level that introduces it, so the six buckets do not overlap and “within HSK 1-3” is the sum of the first three rows.
The word lists behind it
Levels come from the HSK 2012 (HSK 2.0) official word lists — 4995 words containing 2663 unique characters. This tool deliberately uses the historical six-level syllabus so its results remain comparable across runs. These numbers are not a conversion to HSK 3.0 or a statement about the current exam requirements.
| Level | New words | New characters |
|---|---|---|
| HSK 1 | 150 | 174 |
| HSK 2 | 151 | 173 |
| HSK 3 | 297 | 270 |
| HSK 4 | 598 | 447 |
| HSK 5 | 1300 | 621 |
| HSK 6 | 2499 | 978 |
| Total | 4995 | 2663 |
Source: hskhsk.com (Alan Davies) — HSK Official 2012 word lists, retrieved 2026-09-23 from https://hskhsk.com/word-lists.html.
The dataset this page ships is regenerated by scripts/build-hsk-data.mjs. The
generator checks upstream per-level counts against the official totals and stops
regeneration if they exceed the documented tolerance.
Chinese text analysis reference benchmark
These eight original diagnostic samples make the method checkable. They are not a representative reading corpus, graded lessons, or evidence that a learner can comprehend a passage. Results use the same analyzer and bundled word lists as the interactive tool. Version: 2026-10-01.
Download benchmark CSV · Download full results and passages (JSON). The CSV includes all six coverage percentages and a SHA-256 for each passage; the JSON includes the full input and analysis. Original samples and generated results are licensed under CC BY 4.0. Attribute “Fulin Labs Chinese Text Analyzer reference samples” and link to this section.
| Sample | Han | Unique Han | Screening level | HSK 1–6 coverage |
|---|---|---|---|---|
| Everyday sentence | 11 | 10 | HSK 1 | 100.0% |
| Traditional counterpart | 11 | 10 | HSK 1 | 100.0% |
| Commute and reading | 51 | 45 | HSK 3 | 100.0% |
| Technical vocabulary | 27 | 26 | HSK 5 | 100.0% |
| Repetition and one rare character | 10 | 2 | HSK 1 | 90.0% |
| Rare character only | 1 | 1 | Above HSK 6 | 0.0% |
| Chinese with Latin text and emoji | 6 | 6 | HSK 3 | 100.0% |
| No Chinese characters | 0 | 0 | No Han | 0.0% |
Everyday sentence: view passage
我们明天去学校学习中文。
Short everyday sentence; no assigned teaching level.
Traditional counterpart: view passage
我們明天去學校學習中文。
Same sentence in traditional characters; compare the six coverage rows.
Commute and reading: view passage
我每天早上喝一杯咖啡，然后坐地铁去上班。路上我会用手机看一篇中文新闻，遇到不认识的词就先猜意思，晚上再复习一次。
The analyzer’s existing sample passage.
Technical vocabulary: view passage
研究人员用机器学习分析基因序列，发现蛋白质结构发生了变化。
Constructed technical example, not a representative science corpus.
Repetition and one rare character: view passage
我我我我我我我我我龘
Nine familiar occurrences can reach 90% despite one unknown character.
Rare character only: view passage
龘
A Han character outside the bundled lists.
Chinese with Latin text and emoji: view passage
我们学习中文。Hello 2026! 😊
Non-Han characters count toward total length, not HSK coverage.
No Chinese characters: view passage
Hello 2026! 😊
No Han means no assigned level, rather than above HSK 6.
To reproduce a row, copy its exact passage into the input above, leave Your level unset, and download the analysis CSV. Compare the input SHA-256 and six coverage rows. The simplified and traditional pair should have identical coverage. The repetition sample reaches the threshold with one unfamiliar character still present; the no-Chinese sample has no level. Character coverage also cannot tell whether you know a compound word’s meaning or the sentence’s grammar. The 90% cutoff is this tool’s heuristic, not an official HSK reading standard.
What this tool does not do
- It does not add pinyin or translate. A per-character pinyin reading would be wrong on exactly the characters that matter most — 的, 了, 不, 一, 中 all have several readings chosen by context — so this page does not guess at them.
- It does not claim to be a proficiency test. It measures the text, not the reader. Two people reading the same HSK 4 passage can have completely different experiences of it.
- It does not count words the way a Western word counter does. Matched HSK vocabulary is a floor: anything outside the HSK lists — names, places, technical terms — is not matched.
When you want the word-level version instead
This page counts characters against the HSK 2012 lists, which is the right unit for a character-count requirement and the wrong one for "can I read this". Characters are shared between easy and hard words — 银 is not hard, 银河系 is — so character coverage flatters a text.
The HSK Studio text analyzer does the word-level version: it segments the text into words, matches them against the nine-level HSK 3.0 lists, and — because it sits on top of your review deck — marks off the words you have already saved, so what you are left with is the list of words you personally still need. It also exports to Anki, Pleco or CSV. Alongside it there is an HSK level checker, a pinyin converter that reads whole words rather than single characters, and a character frequency counter.
Using the result
Choose your level to list characters outside that historical HSK set, ordered by frequency. Download the complete character study list to review before reading. Repeated familiar characters can raise coverage while unfamiliar characters remain, so use the list and the passage together rather than treating the reported level as a comprehension score.
To read a passage with lookups in place, paste it into HSK Studio (see the product page or the comparison with Pleco, Readibu, and Dong Chinese). To drill the characters the table surfaces, Fulin Flashcard schedules them with spaced repetition. Related reading: how many characters you need to read Chinese, choosing reading material by HSK level, and what the HSK levels actually mean.
Questions
How does this tool decide the HSK level of a text?
It reports the lowest HSK level at which at least 90% of the Chinese characters in your text are covered by the HSK 1-6 character sets. The threshold and the underlying word lists are published on the page, so the number can be reproduced by hand.
Which HSK syllabus is used?
The HSK 2012 (HSK 2.0) official word lists: 4995 words and 2663 unique characters across levels 1-6. This historical six-level dataset is retained for reproducibility; results do not convert to HSK 3.0 or describe current exam requirements.
Does it work with traditional Chinese?
Yes. Traditional characters are read through a traditional-to-simplified table taken from the same HSK 2012 source, so 學習 is counted as 学习 at HSK 1. Results list the characters exactly as you pasted them. A character that is also a simplified character in its own right, such as 著 or 后, is never rewritten.
Does the text I paste get uploaded anywhere?
No. The analysis runs entirely in your browser with JavaScript. The text is never sent to a server and is not stored; the only thing this page remembers between visits is the HSK level you pick under Your level.
Is the character count the same as a word count?
No, and the difference matters in Chinese. This tool counts characters, which is the unit Chinese word-count requirements are usually written in. It also reports how many distinct HSK vocabulary words it can match in the text, which is a floor rather than a true word count, because words outside the HSK lists are not matched.
Why does it separate unique characters from total characters?
Total characters tell you how long a passage is. Unique characters tell you how much you actually need to know to read it. A 1,200-character article often uses only 300-400 distinct characters, which is why re-reading familiar material is cheaper than it looks.
Found a case where the level looks wrong? Send the passage to hello@fulinlabs.com and it can be checked against the published lists.