Perplexity launches Perplexity Computer, a 19‑model workflow that solves tasks end‑to‑end.

Perplexity’s new “Computer” platform is billed as a single‑pane solution that can take a user from concept to deployment without ever leaving the interface. According to the company’s own announcement, the system stitches together 19 distinct AI models, each assigned to a specific sub‑task by an internal routing engine called Opus. Opus matches the nature of the request—whether it’s research, design, code generation, or infrastructure provisioning—to the model best suited for that job, then runs the agents in parallel to keep the workflow moving (Perplexity AI). The result is a “personal computer in 2026” that remembers prior work, offers persistent memory, and provides secure, sandboxed access to files and the web, all built on Perplexity’s infrastructure (Perplexity AI).

The architecture leans heavily on connectors, which the company says number in the hundreds, allowing the platform to pull data from APIs, scrape web content, and interact with cloud services without manual scripting. In practice, a user can clear a to‑do list, push an existing project forward, or spin up a brand‑new side venture with a single command, and the system will automatically allocate the appropriate models to each step of the pipeline (Perplexity AI). A live stream of curated tasks, accessible at perplexity.ai/computer/live, demonstrates the breadth of use cases—from drafting marketing copy to provisioning a Kubernetes cluster—showcasing how the orchestrated agents hand off work without human intervention (Perplexity AI).

Pricing for the service was disclosed in a VentureBeat report, which notes a subscription cost of $200 per month for access to the full suite of 19 models (VentureBeat). The fee positions Perplexity Computer alongside other premium AI‑assistant offerings that target professional users and small teams rather than casual consumers. While the price point is higher than most single‑model chat tools, the company argues that the end‑to‑end capability reduces the need for multiple subscriptions and manual integration, potentially delivering a net cost saving for enterprises that would otherwise stitch together disparate services (VentureBeat).

Industry observers have drawn parallels between Perplexity Computer and earlier “AI‑agent” concepts such as OpenClaw, but note that Perplexity’s approach emphasizes safety and predictability. Ars Technica highlights the platform’s “buttoned‑down, ostensibly safer take” on multi‑agent coordination, suggesting that the internal Opus router enforces strict model selection and sandboxing to mitigate the risk of runaway outputs (Ars Technica). By keeping all interactions within a controlled environment, Perplexity aims to address the trust concerns that have plagued more open‑ended agent frameworks.

The launch arrives at a moment when the market is crowded with specialist AI tools, yet demand for unified, workflow‑centric solutions is rising. TechCrunch points out that Perplexity’s bet reflects a broader industry trend: users increasingly expect a single interface that can handle research, coding, and deployment without hopping between platforms (TechCrunch). If the 19‑model orchestration lives up to its promise, Perplexity Computer could become a reference point for future AI‑augmented productivity suites, setting a new bar for what an “AI‑powered personal computer” looks like in practice.