On Wednesday, Perplexity unveiled Computer, an agentic tool that can orchestrate up to 19 different AI models to execute complex, long-horizon workflows. Available exclusively to Perplexity Max subscribers at $200/month, Computer operates as a cloud-based digital worker that can run tasks for hours or even months.

The system breaks user-defined outcomes into tasks and subtasks, creating specialized sub-agents that leverage different models for optimal performance. Currently, Computer uses Claude Opus 4.6 for core reasoning while deploying Gemini for research, Nano Banana for images, Veo 3.1 for video, Grok for lightweight tasks, and ChatGPT 5.2 for long-context recall.

"Multi-model is the future," Perplexity executives argued, claiming models are specializing rather than commoditizing. The launch of Perplexity Computer follows findings by the company confirming that users frequently switch between models based on task type. According to their reports, during December 2025, users favored Gemini Flash for tasks requiring visual outputs, preferred Claude Sonnet 4.5 for coding, and performed the most medical research using GPT-5.1.

Perplexity Computer is perhaps the first stepping stone in the company's shift toward solutions that cater to enterprise customers after dropping its advertising business last year. Like most companies with a stake in the generative AI business, Perplexity is also chasing the coveted milestone in which it will be able to claim that its tools can significantly contribute to the economy (the company refers to this as enabling "GDP-moving decisions"). The company recently released the Draco benchmark for complex research tasks in which Perplexity's own deep research tool predictably beats all competitors.

As part of its new business strategy, Perplexity is also planning Ask, a developers conference taking place on March 11 in San Francisco, in which the company expects to promote its API.