This research evaluates efficiency tradeoffs for LLM applications across mobile, edge, and cloud environments to address resource constraints on mobile devices.

• Resource constraints create significant challenges for running LLMs directly on mobile devices
• Study implements a comprehensive measurement framework to assess performance metrics across deployment options
• Research examines latency tradeoffs between local processing and cloud connectivity
• Findings help inform optimal deployment strategies for mobile LLM applications

This work provides crucial engineering insights for organizations looking to deliver LLM capabilities on mobile platforms while balancing performance, user experience, and connectivity requirements.

Are We There Yet? A Measurement Study of Efficiency for LLM Applications on Mobile Devices