This research introduces a novel edge-cloud deployment strategy for foundation models that minimizes service delay through intelligent task offloading.

• Proposes a task offloading framework specifically for generative AI models in 6G networks
• Utilizes in-context learning to efficiently handle diverse content generation tasks
• Demonstrates how proper radio resource allocation can optimize foundation model performance
• Provides a blueprint for generative AI as a service in next-generation telecommunications

This engineering advancement is significant for telecom providers and edge computing architects seeking to deploy large language models efficiently across distributed infrastructure while maintaining quality of service.

Generative AI as a Service in 6G Edge-Cloud: Generation Task Offloading by In-context Learning