Count committed characters, not pinyin keystrokes.
Read Chinese characters per minute separately from English WPM. Worked counts explain IME composition, punctuation, corrections and a fixed 60-second round.
These are mode instructions, not a speed band. Every complete guide is available before and after any round, including a zero-input round. A 60-second copy sample does not establish an employer qualification, population percentile or language ability.
Chinese mode reports CPM
Chinese mode reports final committed characters per minute, including punctuation, digits, Latin letters and any spaces actually entered. It does not divide the result by five or report an English-WPM equivalent. A character here is a grapheme segment after canonical normalization, not a physical key or the number of pinyin letters used to select a word. The source passage stays Chinese in either interface. Changing the interface language changes the instructions and guide, not the material being copied.
A full CPM and accuracy example
Suppose 120 characters are committed at the end of 60 seconds and 114 match the source at their respective positions. Final CPM is 120, position-correct CPM is 114, six positions mismatch and positional accuracy is 114/120 × 100 = 95%. There is no English-WPM figure for this round. If a different input method needed more or fewer keystrokes to produce those characters, the final-buffer formula still uses 120; it does not measure that input method’s key efficiency.
Wait for text to be committed
An input method may display a temporary pinyin string or candidate while composition is active. This edition keeps the last committed buffer separately. Only text committed before the deadline enters the result; unfinished composition at the boundary is excluded. For example, a visible candidate for a new word is not added to an already committed 118-character buffer merely because its letters are on screen. If the word is committed in time, its resulting characters become part of the buffer instead.
Copy punctuation and visual line wraps carefully
The source “春天。” contains three grapheme characters: two Han characters and a full stop. Replacing the full stop with a comma changes the third position. A fullwidth punctuation mark and a visually related ASCII mark are not automatically made equal. Screen wrapping is layout, not a newline to type. If you insert or omit a character, later positions can shift, because the accuracy comparison is positional rather than edit distance. Correct the buffer if time remains and inspect where the alignment first changed.
Corrections are allowed within one minute
You may backspace, select and replace text during the round. Deleted or corrected characters are not accumulated as extra output or separately penalized mistakes. Only the final committed buffer is counted, while the time spent correcting remains in the same minute. The untyped remainder of the source is not part of the accuracy denominator. An empty round has zero CPM and no applicable accuracy percentage, rather than an invented 100%. These rules make the correction policy visible rather than treating unlike typing tests as comparable.
Keep the input method and timing conditions visible
Timing starts on the start button, not on the first Chinese character. Pinyin, shape-based methods, prediction, keyboard layout and screen size can change the experience; the page cannot certify how the operating system assisted you. Paste and drop are blocked, but this is not a proctored test or an input-method benchmark. Leaving the page visible is part of the round: backgrounding or a large clock mismatch cancels it, and refresh loses progress. Record consistent conditions for your own practice instead of comparing CPM across unrelated settings.
Choose a small new copy exercise
Look for a concrete issue: selecting an unintended homophone, changing punctuation, omitting a space or losing alignment after a missing character. Make a new short fictional sentence that isolates that issue and copy it carefully before returning to timing. A repeat improvement on a fixed passage may reflect familiarity. Neither CPM nor positional accuracy establishes Chinese proficiency, concentration, an employer cutoff or a job recommendation. Both full language guides remain available without payment, a certificate purchase or another personality test.