Perplexity Launches Hybrid Compute for Mac with Local Privacy Guard

Perplexity introduced hybrid compute for its Mac application, dividing tasks between frontier models in the cloud and a local model on the device. [1] The cloud manages planning, web search, and reasoning, while the local model processes sensitive information, private files, and on-device actions. [1] The feature is available to Perplexity Pro, Max, and Enterprise subscribers using Apple silicon Macs with macOS 15 or later and at least 24GB of unified memory. [1][3] Users can download a local model in one click without a separate runtime setup, and local work consumes no cloud credits. [1][3]

The Mac release inverts the design of Perplexity's recently launched local compute mode on the NVIDIA DGX Spark, which defaults to local hardware and escalates to the cloud. By orchestrating from the cloud downward, the system enables users to trigger tasks remotely via iPhone while local steps execute on their Mac. To support the boundary between environments, Perplexity open-sourced the classifier that powers its on-device privacy gate. What the sources don't address: Whether the open-sourced classifier will see adoption by third-party developers building similar hybrid AI agents.