Abstract
The integration of Large Language Models (LLMs) into Virtual Reality (VR)games marks a paradigm shift in the design of immersive, adaptive, andintelligent digital experiences. This paper presents a comprehensive review ofrecent research at the intersection of LLMs and VR, examining how these modelsare transforming narrative generation, non-player character (NPC) interactions,accessibility, personalization, and game mastering. Drawing from an analysis of62 peer reviewed studies published between 2018 and 2025, we identify keyapplication domains ranging from emotionally intelligent NPCs and procedurallygenerated storytelling to AI-driven adaptive systems and inclusive gameplayinterfaces. We also address the major challenges facing this convergence,including real-time performance constraints, memory limitations, ethical risks,and scalability barriers. Our findings highlight that while LLMs significantlyenhance realism, creativity, and user engagement in VR environments, theireffective deployment requires robust design strategies that integratemultimodal interaction, hybrid AI architectures, and ethical safeguards. Thepaper concludes by outlining future research directions in multimodal AI,affective computing, reinforcement learning, and open-source development,aiming to guide the responsible advancement of intelligent and inclusive VRsystems.