Perplexity enables fully local AI Computer workflows on Apple Silicon Macs

Perplexity released an update on October 20, 2026, allowing users with Apple Silicon Macs equipped with at least 36GB of RAM to run the entire Computer application locally, including its core harness and orchestrator components, without sending data to the cloud.

Users can choose to run Computer locally with either Qwen 3.6 or PPLX Pro, the latter being Perplexity's post-trained variant of Qwen 3.6 fine-tuned for Computer use. They can also authorize the software to call on more than 15 cloud-based frontier models when needed, with all code or tool execution happening in an isolated sandbox environment.

Perplexity said the feature is designed to give users more control over sensitive data and credit spend while reducing costs. Perplexity Pro, Max and Enterprise subscribers can start using Computer locally today by downloading the latest Mac app and following in-app instructions to install a local model.