2 min read AI-generated

Perplexity Computer – A Digital Worker with 19 AI Models

Copy article as Markdown

Perplexity has launched Computer, a platform that orchestrates 19 AI models and autonomously works through entire projects. Claude Opus 4.6 provides the core engine.

Featured image for "Perplexity Computer – A Digital Worker with 19 AI Models"

Perplexity was previously known mainly as a clever search engine. That’s history now. On February 25th, the company unveiled Perplexity Computer – and with it, a fairly radical pivot.

What Perplexity Computer Is

Computer is not a chatbot. It’s a digital worker that executes complete projects from start to finish. You describe the desired outcome, and Computer handles the rest: researching, creating documents, writing code, processing data, generating images, connecting services.

The special part: under the hood, 19 different AI models work together, each optimized for its strengths. Claude Opus 4.6 from Anthropic serves as the central reasoning engine – the brain that plans and coordinates the work. Additionally, GPT-5.2, Gemini, and Grok are used, among others.

How It Works

You give Computer a task, and the system autonomously breaks it down into subtasks. For each one, a specialized sub-agent is launched. These agents work in parallel – one researches while another writes code and a third creates designs.

The crucial difference from other AI tools: Computer works asynchronously. You don’t have to sit there and keep prompting. The system can run in the background for hours or even days until the project is finished.

What It Costs

Computer is currently only available in the “Perplexity Max” subscription – for $200 per month. That includes 10,000 credits, plus a welcome bonus of 20,000 credits. Pro and Enterprise access is planned after a testing phase.

Why This Is Relevant

The market for AI agents is being contested from all sides. OpenAI has the ChatGPT Agent, Anthropic has Cowork, Google is working on Gemini automations. Perplexity takes a different path: instead of relying on one model, they orchestrate all major models simultaneously.

Whether this works in practice remains to be seen. The idea of having 19 specialized models collaborate sounds elegant – but also error-prone. One thing is certain: the race for the best digital assistant is accelerating massively right now.

Sources: Perplexity Blog, TechCrunch, Business Standard