Perplexity, a company developing AI search and its own browser, has introduced a new product, the Perplexity Computer. It is an AI system that takes your goal, breaks it down into steps, and brings it to a result, performing actions in interfaces the way a human would.
Perplexity, a company developing AI search and its own browser, has introduced a new product, the Perplexity Computer. It is an AI system that takes your goal, breaks it down into steps, and brings it to a result, performing actions in interfaces the way a human would.
As Perplexity writes in its release on the website, the user describes the desired result, after which Computer builds a plan of subtasks and launches separate performers for different parts of the work. One can collect information on the network, another can prepare a document in parallel, and a third can process data or make requests to connected services via API. The company claims that the system is able to conduct such processes for a long time, without requiring constant control, and involves the user only when it is really necessary.
Perplexity emphasizes that each task runs in an isolated environment with a real browser and file system. If a problem occurs during execution, Computer can create additional “subtasks” to find missing data, clarify instructions, prepare code, or gather needed materials.
Another claimed feature is the combination of multiple models within a single process. Perplexity calls Opus 4.6 the “thinking core,” and for individual stages it connects other models: Gemini for deeper research, Grok for quick tasks, Nano Banana for images, Veo 3.1 for video, and ChatGPT 5.2 for long context and search. The user can also choose models for specific subtasks.
Perplexity is developing an ecosystem around AI search, including the Comet browser and Comet Assistant. Perplexity Computer is initially available to Perplexity Max subscribers, and is expected to be available to Enterprise Max users later.
Previously, dev.ua wrote about how Perplexity is abandoning advertising amid debates about trust in AI.
