With the advancement of AI technology, virtual streamers are increasingly being utilized in live streaming commerce to reduce costs and enhance efficiency. However, there remains a lack of clear guidance on how companies should rationally allocate resources between identity-based and task-oriented virtual streamers. This study aims to compare the effectiveness of identity-based and task-oriented virtual streamers in recommending different product types. Based on construal level theory, this study develops a theoretical model to examine the interactive effects of virtual streamer type and product type on consumers’ purchase intentions, along with the underlying influencing mechanisms. Through a 2 (virtual streamer type: task-oriented vs. identity-based) × 2 (product type: utilitarian vs. hedonic) between-subjects experimental design, data from 302 valid participants were collected to test the theoretical model. The results indicate that identity-based virtual streamers are more effective than task-oriented virtual streamers in recommending hedonic products, while no significant difference was found for utilitarian products. Moreover, psychological distance and trust chaining mediated this effect. This study enriches the literature on virtual streamers and offers practical guidance for practitioners in the live streaming commerce industry.