# Perplexity Launches Local-First AI Mode for Mac Users New offline capability enables on-device task execution and private file processing for Mac users. - **Source:** [9to5Mac](https://brifing.si/source/59447403-ce5b-427c-8132-8614f549929a) - **Published:** 2026-10-20 - **Topic:** [Data & AI](https://brifing.si/c/data-ai) - **Original article:** - **On Brifing:** https://brifing.si/s/c11f0b99-e23d-4a67-b1ec-7cb385a46ab6 ## Local-First AI Execution Perplexity has introduced a local-first mode for its Mac application, allowing the Computer agent to function entirely on-device. This development moves beyond the previous Hybrid Compute model, which debuted in September. By running both the AI model and the task-directing agent locally, users can now perform complex operations without any active internet connection. ## Privacy and Data Control The primary advantage of this local-first approach is the ability to process sensitive information without cloud transmission. Planning, reasoning, and file processing occur exclusively on the Mac hardware. Files are only uploaded to the cloud if the user explicitly grants permission, ensuring that private data remains under the user's direct control at all times. ## Native Application Integration With user authorization, the Perplexity Computer agent can interact directly with native Mac applications. This includes reading and processing data from Notes, Messages, Mail, and Numbers. For example, the agent can scan recent emails and messages to identify upcoming deadlines, subsequently organizing this information into a structured action list within the Notes app. ## Hardware Requirements To utilize these local-first capabilities effectively, users must possess an Apple silicon Mac. The company specifies a minimum requirement of 36GB of unified memory for an optimal experience. This hardware threshold is necessary to handle the computational demands of running the AI models and the agent simultaneously on the local machine's internal architecture. ## Model Selection Options The feature launches with support for Qwen 3.6 and PPLX Pro as the primary local model options. Users are required to download and configure these models directly through the Perplexity Mac application. This setup process allows for a tailored experience, enabling users to choose the model that best fits their specific local processing requirements. ## Hybrid Compute Coexistence The new local-first mode does not replace existing features. Hybrid Compute remains available for tasks that necessitate real-time information, browser access, or connectivity with external applications. Users retain the flexibility to authorize cloud access whenever a specific task requires external data sources, maintaining a balance between local privacy and cloud-based intelligence capabilities. ## Subscription Tier Access Access to this local-first functionality is currently restricted to specific user segments. Only Perplexity Pro, Max, and Enterprise subscribers can utilize the on-device processing features. This rollout strategy targets power users who require advanced AI integration within their local workflows, particularly those handling sensitive data that must remain off the public cloud. ## Strategic Shift in AI This update represents a significant shift toward edge computing for AI agents. By enabling offline task execution, Perplexity addresses critical concerns regarding data security and latency. This transition allows professional users to leverage AI for sensitive document management and internal communication analysis, marking a new phase in how AI agents integrate with personal computing. ## Sources covering this story - https://9to5mac.com/2026/10/20/perplexitys-computer-can-now-run-entirely-on-your-mac-even-offline/ — --- _Summary written by Brifing from the original reporting. Copyright in the original article belongs to its publisher; read the full piece at the link above._