Perplexity enables fully local AI workflows on Apple Silicon Macs

Users can now execute the full Perplexity Computer experience directly on their own hardware, eliminating cloud dependency for specific tasks. This capability requires a recent Apple Silicon Mac equipped with at least 36GB of RAM.

Local execution capabilities expand

The update follows the September introduction of Hybrid Compute, which allowed splitting workloads between frontier and local models. Perplexity has now extended this functionality to enable running the platform entirely on-device, including the harness and orchestrator components.

The company positions this feature as beneficial for cost-conscious users and those prioritizing data privacy. By accessing local applications like Notes and Mail, users can generate priority to-do lists without sending email data to the cloud. "As more demanding workflows become practical to run locally, the machines people already own will take on a larger share of their AI work," Perplexity stated. "This gives users more control over sensitive data and credit spend."

Model selection and security protocols

Individuals can choose between two models for local operation: Qwen 3.6 or PPLX Pro. The latter is a Perplexity-developed variant of Qwen 3.6, fine-tuned through proprietary post-training for Computer use.

If a task benefits from external processing, users may authorize the software to access one of more than 15 frontier models. All code and tool executions occur within an isolated sandbox environment. "These same security protocols persist in the cloud version of Perplexity Computer," the company noted.

Subscription access and setup

Access to local Computer functionality begins today for Perplexity Pro, Max, and Enterprise subscribers. Users must download and install the latest version of the Perplexity Mac app, then follow in-app instructions to deploy a local model on their machine.