We study the presence of heteronormative bi002 ases or prejudice against inter-racial marriages in large language models by performing controlled name-replacement experiments for the task of relationship prediction. We show that models are less likely to predict romantic relationships for (a) same-gender character pairs than different-gender pairs; and (b) inter/intra-racial character pairs involving Asian names as compared to Black, Hispanic, or White names. We closely examine the contextualized embed012 dings of first names and find that gender for Asian names is less discernable than non-Asian names. Additionally, we delve into the social implications of our findings, underlining the need to prioritize the development of inclusive and equitable technology.