Perplexity Computer is an artificial intelligence system developed by Perplexity to function as a "general purpose digital worker." The system is designed to create and execute complex, long-running workflows by integrating multiple advanced AI models into a coordinated platform. It moves beyond single-prompt interactions by breaking down large user-defined goals into subtasks and autonomously managing their execution over extended periods. ^[1]

History and Launch

Perplexity Computer was announced on February 25, 2026, as a natural evolution of Perplexity's previous products. It builds upon the company's foundational AI search engine and its AI-native Comet Browser, integrating their capabilities with deeper research functions and persistent task management to form a comprehensive workflow system. ^[1]​

Upon launch, Perplexity Computer was made available to subscribers of the company's "Max" plan. Perplexity has also announced plans to offer Enterprise access in the future. ^[1]​

Core Capabilities

Perplexity Computer is designed to manage entire workflows rather than responding to single prompts. Its primary capabilities include autonomous task management, long-duration operation, and self-correction. ^[1]

• Workflow Coordination: The system can take a high-level goal and create a detailed plan, breaking it down into smaller tasks and subtasks for execution.
• Autonomous Task Management: It deploys specialized sub-agents to handle different parts of a workflow. If a task fails, the system can autonomously generate new agents to troubleshoot the problem, conduct research, or write code to resolve the issue.
• Long-Duration and Asynchronous Operation: Workflows can run for hours or even months. The system operates asynchronously, allowing users to initiate tasks and return later or run multiple workflows in parallel.
• Isolated Compute Environment: Each task is executed within a secure, isolated compute environment equipped with essential tools, including a web browser, a filesystem, and the ability to make API calls.

These features allow the system to handle complex, multi-step projects that are beyond the scope of traditional chatbots or single-task agents. ^[1]

Technical Architecture

Perplexity Computer is built on a "multi-model orchestration" framework. Instead of depending on a single AI model family, it routes different types of tasks to the most suitable frontier model available, leveraging the specialized strengths of each. ^[1]​

The models utilized at launch include:

• Core Reasoning Engine: Opus 4.6
• Research Tasks: Gemini
• Image Generation: Nano Banana
• Video Generation: Veo 3.1
• Lightweight & Speed-Sensitive Tasks: Grok
• Long Context Recall & Search: ChatGPT 5.2

This multi-model approach allows the system to flexibly apply the best available technology to each component of a complex task. ^[1]​

Use Cases and Examples

The system is designed to handle complex, end-to-end projects that require research, planning, execution, and iteration. It can be used for tasks ranging from market analysis and software development to creative project management and personal planning. The autonomous and long-running nature of the platform makes it suitable for goals that would typically require significant human oversight and coordination. ^[1]

Vacation Planning

A user could task Perplexity Computer with a high-level goal like: "Plan a 10-day family vacation to Italy in July for two adults and two children, with a budget of $8,000, focusing on historical sites and family-friendly activities." The system would break this down into a multi-step workflow:

1. Research: It would first research visa requirements, weather conditions in July, and major holidays or events. It would then identify family-friendly cities and historical sites.
2. Itinerary Creation: Based on the research, it would draft a potential itinerary, suggesting a route (e.g., Rome, Florence, and Venice), allocating days to each city, and proposing specific activities.
3. Booking and Logistics: The system would then search for flights and accommodations that fit the budget and user preferences, presenting the top options. It could also research train schedules between cities and options for local transport.
4. Refinement: The user could review the draft plan and provide feedback, such as "spend less time in Florence and add a day trip to the countryside." The Computer would then autonomously adjust the itinerary and bookings.

This entire process could run over several days, with the system providing updates and requesting user input only when necessary. ^[1]

Future Developments

Perplexity plans to expand the platform's capabilities and create an ecosystem around Perplexity Computer. In addition to the planned Enterprise version, the company's vision includes the development of a developer API. This API would allow third-party developers to integrate their own tools and services into the Computer's workflows. Furthermore, there is a long-term vision for an "App Store for Computers," where users could access and deploy pre-built, specialized workflows created by partners and the community, further extending the system's utility. ^[1]