That size in bytes is based on the max size in UTF-8 and UTF-16. Codepoints below U+1100 are counted as one "character" by twitter and will need at most 2 bytes. Codepoints above it are counted as two "characters" by twitter and will need at most 4 bytes. Therefore 560 bytes, and it supports all languages.
Side note, this is more pessimistic than it needs to be, if you're willing to transcode. The larger codepoints fit into 20-21 bits, and the smaller ones fit into 12-13 bits.