LLMs on XR (LoXR): Performance Evaluation of LLMs Executed Locally on Extended Reality Devices
We evaluate 17 LLMs across four XR devices—Magic Leap 2, Meta Quest 3, Vivo X100s Pro, and Apple Vision Pro—assessing performance on key metrics: consistency, processing speed, and battery consumption. Our experimental setup examines 68 model-device pairs under varying string lengths, batch sizes, and thread counts, providing insights into trade-offs for real-time XR applications. Our findings offer guidance for optimizing LLM deployment on XR devices and establish a foundation for future research in this rapidly evolving field.
Authors
- Donggang Jia (ORCID: https://orcid.org/0000-0002-1358-8718)
- Xinyu Liu
- Ivan Viola (ORCID: https://orcid.org/0000-0003-4248-6574)
- Alexandre Kouyoumdjian (ORCID: https://orcid.org/0009-0005-0915-369X)
- Dawar Khan (ORCID: https://orcid.org/0000-0001-5864-1888)
- Omar Mena
Institutions
- King Abdullah University of Science and Technology (SA)
Publication Details
- Published
- 2025-03-08
- DOI
- https://doi.org/10.1109/vrw66409.2025.00252
- Primary Topic
- Advanced Data Storage Technologies
- Type
- article
- Field-Weighted Citation Impact
- 0.00