When selecting Uzbek or Kazakh languages (possibly others), the language variant (i.e. cyrillic or latin script) does not seem to be respected.
For example:
curl --header "Accept-Language: uz-cyrl" https://uz.wikipedia.org/api/rest_v1/page/html/Yer
Requesting the cyrillic version of [[uz:Yer]] returns latin text.
Interestingly, when requesting the mobile-html endpoint instead of the plain html endpoint, the page title is correctly Cyrillic, but the rest of the article is not:
curl --header "Accept-Language: uz-cyrl" https://uz.wikipedia.org/api/rest_v1/page/mobile-html/Yer