Perplexity said on the 5th that it unveiled Perplexity Computer, which integrates different state-of-the-art artificial intelligence (AI) models into a single system.
Perplexity Computer implemented a work system that creates complex workflows over extended periods by combining multiple models to fit the nature of each task rather than being tied to any specific AI model. Instead of one model handling all tasks, it boosts overall performance by assigning the best-performing model to the right role.
It uses Anthropic's Claude Opus 4.6 as the core reasoning engine. For in-depth research it deploys Google Gemini, for image generation Nano Banana, for video Vio 3.1, for lightweight tasks xAI's Grok, and for long-context handling and search OpenAI's ChatGPT 5.2, intelligently assigning models optimized for each task.
When a user describes the desired end result, it breaks that down into tasks and subtasks, spawns sub-agents to handle each, and runs them in parallel. For example, while one AI agent writes a document, another agent gathers data or conducts analysis. It performs work by directly manipulating the actual software stack, including web research, document writing, data processing, and application programming interface (API) calls.
Within an isolated safe environment, the system creates and edits real files, browses the web, and uses external tools. If a problem arises during work, it spawns additional sub-agents to attempt a fix and asks for user confirmation only when necessary. By default, it resolves issues autonomously.
With this launch, Perplexity said it has entered "a new phase where AI becomes the computer." This means AI moves beyond merely using tools on behalf of users to becoming the entity that runs the entire computing environment.
Perplexity Computer is available to Perplexity Max subscribers and will be expanded to Enterprise Max users in the future.