Counts as you type. "One character" means different things depending on how you count, so several measures are shown side by side.
0Characters (graphemes)
0Code points
0Lines
0Words
0UTF-8 bytes
0Full-width alphanumerics
0Half-width alphanumerics
0Full-width kana
0Half-width spaces
0CRLF line endings
0Surrogate-pair surplus
X (Twitter) limit check
—
X weights Latin letters and digits as 1 and everything else (CJK, emoji) as 2, then compares against 280. A plain character count cannot tell you whether a post fits.
Why "character count" disagrees
Take 👨👩👧 as an example: it is several code points joined together to look like one character.
Grapheme — the unit a human perceives as one character. This is the right measure for copy and manuscript checks
Code point — one Unicode character; combined emoji count as several
UTF-16 units / bytes — units for programs and file sizes; CJK is 3 bytes per character in UTF-8
If "surrogate-pair surplus" is greater than 0, the text contains emoji or CJK extension characters.