Perplexity计算机将AI任务在本地与云端模型间分配

Perplexity AI is developing a new capability for its Perplexity Computer platform that will allow AI tasks to be divided between on-device processing and cloud-based models. The approach reflects a broader shift toward hybrid computing, where performance, privacy, and efficiency are balanced across different environments. Rather than relying entirely on remote servers, the system is designed to determine which parts of a task can run locally and which require the scale of cloud infrastructure. A HYBRID MODEL FOR AI WORKLOADS The core idea behind the update is to intelligently distribute workloads. Simpler or privacy-sensitive tasks—such as summarizing personal documents or analyzing local files—can be handled directly on the user’s device. …