This current study reports three multilingual Chinese students’ audience design strategies on a populated Social Networking Site (SNS), WeChat. Considering the importance of audiences in shaping multilingual speakers’ language choice (Bell, A. (1984). Language style as audience design. Language in Society, 13(2), 145–204) and the potential hazard of conflated audiences in social media, as well as the comparative lack of research on Wechat which has different technical affordances from the well-researched SNS, Facebook, this article aims to shed new light on how multilingual speakers harness their linguistic repertoire to cope with the “context collapse” (Marwick, A. E., & Boyd, D. (2011). I tweet honestly, I tweet passionately: Twitter users, context collapse, and the imagined audience. New Media & Society, 13(1), 114–133) in semi-public SNS. Data collection and analysis follow an online ethnography method consisting of 588 initial posts from participants’ WeChat Moments during 1 year and semi-structured interviews. Theoretically informed by Androutsopoulos’s (2014b; Languaging when contexts collapse: Audience design in social networking. Discourse, Context & Media, 4–5, 62–73) audience specification framework, the findings reveal that although the default language in both online and offline interactions is Chinese, Wechat users have a very high sensitivity to different patterns of Chinese–English code-switching and grammatical and lexical choices in English. Meanwhile, Chinese multilingual speakers usually developed highly nuanced audience design strategies to target or partition specific groups of audience, which are far more complicated than the audience design strategies found in previous research on Facebook.