chinese models feel strong in japan — kanji. but outside language? maybe ... max sonnet 4.5 level.

do benchmarks reflect that gap in english region?