Perplexity's New 'Computer' Coordinates 19 AI Models So You Don't Have To — For $200 a Month

Perplexity Computer multi-model AI agent orchestration system

Perplexity has launched "Computer," a multi-model AI agent that coordinates 19 different AI models to handle complex tasks autonomously. It's available now on the company's $200/month Perplexity Max tier, and it represents the clearest articulation yet of a thesis that the future of AI isn't one model to rule them all — it's orchestration.

What Computer Actually Does

Describe what you want — "Plan a weeklong trip to Japan, find flights under $1,200, and build a full itinerary with restaurant reservations" — and Computer breaks the project into components, assigns each to the right AI model, and works on it in the background. It can even spawn subagents to handle specific sub-problems.

The key word here is "background." Unlike chatbot interactions where you sit there watching tokens stream, Computer takes your request and goes away to work on it. You come back later to a finished product. It's closer to delegating to an employee than having a conversation with a chatbot.

The Model Orchestra

Here's what's actually running under the hood:

  • Anthropic's Claude Opus 4.6 — central reasoning engine
  • Google's Gemini — deep research queries
  • Google's Nano Banana — image generation
  • Google's Veo 3.1 — video generation
  • xAI's Grok — lightweight tasks
  • OpenAI's GPT-5.2 — long-context recall and expansive web search

Plus 13 other models handling specialized tasks. The idea is that each model has strengths, and rather than forcing one model to do everything badly, you route each subtask to the model best suited for it.

The $200 Question

At $200/month, Computer is firmly in the "this better save me real money" category. Perplexity is essentially saying: if you're already paying for multiple AI subscriptions, why not pay us once and let us handle the routing?

It's a reasonable pitch for power users and small businesses. Whether it justifies the premium over individual model subscriptions depends entirely on how well the orchestration works in practice. Bad routing — sending a complex reasoning task to a lightweight model, or using an expensive model for a simple lookup — would burn through the value proposition quickly.

The Bigger Bet

What makes Computer interesting isn't the product itself — it's the thesis behind it. Perplexity is betting that AI models are specializing, not converging. That the future isn't one god-model that does everything, but an ecosystem of specialized models that need an intelligent router.

This runs counter to the narrative from OpenAI, Google, and Anthropic, all of which are building increasingly general-purpose models. Perplexity is saying: let them compete on individual model quality, and we'll be the layer that sits on top and makes the best use of all of them.

It's a smart hedge. If the general-purpose models converge into commodities, the orchestration layer becomes the valuable piece. If they don't converge and specialization persists, the orchestration layer is essential.

Cloud-Only Has Its Advantages

Computer runs entirely in the cloud, which sidesteps some of the security headaches plaguing other agentic AI tools. There's no local browser automation to exploit, no file system access to abuse. The tradeoff is that you're trusting Perplexity with your data and your task descriptions, but that's a more contained risk than giving an AI agent access to your actual computer.

Whether the orchestration thesis plays out remains to be seen. But at minimum, Perplexity has identified a real gap in the market: most people don't want to become AI model experts. They just want their work done.