Evergreen Tools
|
← Back to AI Tools

Perplexity Computer Runs Fully Local on Mac

On 20 October 2026 Perplexity made Computer run fully local on Mac: the model, the harness and task execution all stay on the device, and with permission it can work across local files and Mac apps such as Notes and Mail without sending personal data to the cloud

Tool Interface

Interactive tool will be available soon

Features

  • ✓ Model, harness and task execution all run on the device rather than in the cloud
  • ✓ With permission it works across local files and Mac apps like Notes and Mail, keeping personal data on the machine
  • ✓ The Mac app offers two paths: Hybrid Compute starting in the cloud, or a fully local run starting on the Mac
  • ✓ Local model options are Qwen 3.6 and PPLX Pro, the latter a post-trained version of Qwen
  • ✓ Available to Pro, Max and Enterprise subscribers on supported Apple silicon Macs, with 36GB of memory or more recommended

How to Use

  1. Confirm you are on a supported Apple silicon Mac - Perplexity cites the new M6 Mac mini and M5 Pro machines
  2. Sign in to a Pro, Max or Enterprise account in the Perplexity Mac app
  3. Choose to run Computer locally rather than using Hybrid Compute
  4. Grant access to local files and apps such as Notes and Mail when prompted, then start the task

FAQ

Does fully local mode still use the network?

Perplexity states the model, harness and task execution all run on the device, so personal data from local files stays on the machine. Anything requiring live retrieval still depends on the capabilities and permissions you enable.

What hardware does it need?

A supported Apple silicon Mac. Perplexity recommends 36GB of memory or more for running Computer locally.

How is this different from Hybrid Compute?

Hybrid Compute begins in the cloud and coordinates research and reasoning with local work on private files. Running locally begins on the Mac and keeps planning, reasoning and file processing on the device.

Which model runs locally?

Perplexity says it launches with Qwen 3.6 and PPLX Pro, which is a post-trained version of Qwen.

Can free users run it?

Perplexity lists it as available to Pro, Max and Enterprise subscribers, so the free tier is not included.