Explained: What is Perplexity Computer, what it can do, and more

Perplexity Computer

• Perplexity launches Computer, an AI model orchestration feature
• Computer routes tasks to specialist AI models for efficiency
• Max subscribers get feature now; Enterprise Max access soon

Did our AI summary help?

Perplexity has introduced a new feature called Computer that allows multiple AI models to collaborate on the same workflow. Announced by CEO Aravind Srinivas, the system dynamically taps different frontier models depending on the job at hand. It is live for Perplexity Max subscribers starting today and will roll out to Enterprise Max users soon.

Despite the name, Computer is not hardware. It is an orchestration layer designed to coordinate models behind the scenes.

What is Perplexity Computer?

Perplexity AI Computer is built as a model orchestration system. Instead of forcing one large language model to handle everything from research to recall, it routes subtasks to specialist engines that are better suited to each requirement.

At its core sits Opus 4.6, which acts as the primary reasoning engine. When a task demands something specific, the system hands it off. For deeper research, it uses Gemini from Google, which can create sub-agents to break down complex queries. For faster, lightweight responses, it calls on Grok from xAI. When long-context recall or wide-ranging searches are needed, it deploys ChatGPT 5.2 from OpenAI.

The philosophy is straightforward. Rather than stretching one system across every use case, use the right model for each component of the workflow.

Why splitting tasks across models matters

Perplexity’s approach attempts to challenge the notion that frontier AI models are becoming interchangeable commodities. Its argument is that models are specialising, not converging.

Some engines excel at deep reasoning and structured research. Others are optimised for speed. Some handle long context windows and conversational memory more effectively. By letting each system play to its strengths, the overall output becomes more capable.

The name Computer is a historical nod. In the 1700s, human “computers” divided complex calculations into smaller pieces and worked collaboratively. Perplexity is applying the same division-of-labour principle to AI models.

The result is an experience that feels less like interacting with a single chatbot and more like coordinating a team.

Do users get control? Yes. While Computer assigns default routing logic, it is model agnostic and allows users to choose which models handle their subtasks.

This level of control becomes important as token budgets and usage costs grow more significant in professional settings. If one model consumes more credits for simple tasks, users can direct those jobs to a cheaper or faster alternative. The flexibility also means Perplexity can swap out backend engines as newer models emerge, without fundamentally changing how the feature works.

What this means for subscribers

Computer is currently available to Max subscribers of Perplexity AI, with Enterprise Max access expected soon.

Instead of asking which single AI model is best, the focus moves to how effectively multiple models can be orchestrated together. If the system quietly assigns faster models to simple searches and deeper reasoning engines to complex research, workflows should feel smoother without constant manual switching.