Why emoji can change your character count
One emoji, seven UTF-16 units. Find out which count your draft actually needs.
An emoji can look like one character and still occupy several units in a text field's internal representation. That is why two counters can disagree without either losing your text.
CozyToolkit's word counter counts Unicode grapheme clusters: groups that usually correspond to what you see as one character. It counts a joined emoji such as ๐ฉ๐ฝโ๐ป as one character, rather than counting its component parts separately.
Try one short draft
Open the sample text file, copy its contents, and paste them into Your text. There is one blank line between the two lines of text and no final newline:
Cafรฉ ๐ฉ๐ฝโ๐ป
Ready to share?
The result is 4 words, 23 characters, 18 without spaces and 2 paragraphs. The emoji is a character, but it is not counted as a word. Punctuation counts towards characters too.
If your editor adds a newline at the end when you copy, the total character count changes. Start with the downloaded file when checking the exact numbers. Clear removes the draft; Undo clear brings it back if you want to compare again.
One visible character can have several representations
These four samples all count as one character in this tool:
| Sample | Characters here | Unicode code points | JavaScript UTF-16 units | UTF-8 bytes |
|---|---|---|---|---|
A |
1 | 1 | 1 | 1 |
รฉ |
1 | 1 | 1 | 2 |
eฬ |
1 | 2 | 2 | 3 |
๐ฉ๐ฝโ๐ป |
1 | 4 | 7 | 15 |
The two accented letters look alike. The first stores the accented letter as one code point; the second stores a plain letter followed by a combining accent. The emoji combines a person, a skin-tone modifier, a joiner and a laptop into one displayed symbol.
The tool uses the browser's Intl.Segmenter for character, word and sentence boundaries. It does not simply split words at spaces, which matters for languages whose words are not separated that way. Language rules and browser Unicode data can still affect word and sentence segmentation. A word count is not a count of AI tokens.
Download the exact strings and their counts to inspect the different representations. The byte and UTF-16 counts in that file are reference measurements; they are not extra controls in the word counter.
Without spaces also excludes line breaks
In the sample, three spaces and two newline characters account for the difference between 23 and 18. Without spaces excludes whitespace, including tabs and line breaks, rather than removing just ordinary spaces. Blank lines separate paragraphs; a single line break does not start another paragraph in this counter.
This means a caption can have the same number of words after you reformat it, while its character and paragraph counts change. Copy the text as you intend to publish it, including its line breaks.
Check the destination's own limit
A social platform, form or API can count text differently. Its limit may use bytes, UTF-16 units or platform-specific rules rather than the characters shown here. Use this counter to edit your draft, then check the final destination before publishing. A lower count here does not override a rejection there.
Reading time is a rough estimate at 200 words per minute, so this short draft shows less than one minute. It does not measure your audience's reading speed. Counting happens on your device and needs no AI call.
The results record reproduces these samples with the tool's counting function on 2 November 2026. The draft was also checked in the browser UI. The examples do not test social-network publishing limits or every language's word boundaries.