Perplexity Launches Hybrid Compute for Mac, Splitting Tasks Between Cloud and Local Models

Perplexity on September 1, 2026 introduced hybrid compute for the Perplexity Mac app, a mode that divides each Perplexity Computer task between frontier models in the cloud and a local model running on the Mac itself. [1]

The feature runs on any Apple silicon Mac with macOS 15 or later and at least 24GB of unified memory. [1] It launches with three local models: Gemma 4 E4B, Qwen3.6 35B-A3B, and a Perplexity model post-trained for Perplexity Computer. [1]

Under hybrid compute, the cloud handles frontier reasoning, web search, and planning, while the local model on the Mac processes private files, sensitive information, and on-device actions. [1]

The system is built around a privacy gate on the Mac that controls what may leave the machine. [1] Before information from a protected file reaches the cloud, the gate can mask sensitive details, keep the content local, refuse the action, or ask the user for consent. [1]

By introducing hybrid compute, Perplexity addresses a central friction point in enterprise AI adoption: balancing the superior reasoning power of cloud-based frontier models with the strict data privacy requirements of local processing. The strategy directly leverages the unified memory architecture of Apple silicon, positioning the Mac mini and Mac Studio as localized command centers for automated tasks. Unlike purely cloud-reliant agents, this dual-model approach mitigates the risk of leaking proprietary files. What the sources don't address: Whether Perplexity intends to bring this hybrid processing architecture to Windows PCs with comparable local AI capabilities.