Perplexity and Nvidia Launch Local AI Agent Platform, Moving Workloads Off Cloud

Perplexity and Nvidia announced Portable Computer on Monday, a local AI agent platform that runs entirely on hardware users already own—starting with Nvidia's DGX Spark desktop supercomputer and Linux machines equipped with Nvidia RTX GPUs. The launch represents one of the most serious attempts yet to move AI agent workloads off the cloud and onto local devices, where the model, the user's files, and the work itself can all remain private, with no billing charges accumulating.

The platform packages a complete AI stack into a single application: the model, the agent harness (the scaffolding of prompts, tools, and orchestration logic), the inference engine, connectors to services like Gmail and GitHub, and an OS-level security sandbox. Users can start with either Qwen 3.8 27B or PPLX 27B, a version Perplexity has post-trained on its own harness, with Nvidia's Nemotron 3.5 Lightning coming soon. The system runs on any RTX GPU with at least 24GB of VRAM—roughly a GeForce RTX 3090 or newer—and launches today for Pro, Max, Enterprise Pro, and Enterprise Max subscribers on Linux, with Windows support following in September. By default, every task starts on the device; the system asks permission before sending any individual step to a more powerful frontier model in the cloud.

During a press briefing Monday, Nate, Perplexity's vice president of engineering for infrastructure and enterprise, demonstrated the platform reviewing a folder of 1099s and investment documents—the kind of sensitive financial material many users would hesitate to upload to a cloud service. Running a 27-billion-parameter Qwen model at full GPU utilization on a DGX Spark, the agent reviewed each document and flagged cases where unnecessary fees were being paid. The interface element that normally displays a running tally of cloud credits, Nate noted, "is just parked at zero, because all of this is happening on the device." A second demo showed the hybrid side of the product: the agent analyzed a CSV of user funnel data locally, then pushed the finished analysis to a Slack channel using Perplexity's connector ecosystem, proving that local-first does not mean disconnected.

Perplexity's research paper, published alongside the launch, argues that effective local agents require the model and the agent harness to be designed together rather than assembled from general-purpose components. The core insight is that general-purpose harnesses assume a frontier model capable of absorbing enormous contexts, navigating sprawling tool surfaces, and planning over long horizons—demands that small local models cannot meet. Perplexity found empirically that although Qwen 3.8 27B advertises a 260,000-token context window, it begins to struggle beyond 100,000 tokens. The company built a deliberately minimal harness: a succinct system prompt, a small set of core tools, and capabilities that load and unload as on-demand "skills" rather than sitting permanently in context. It converted popular connectors like Gmail and GitHub from token-hungry MCP servers into compact command-line tools, added self-verification hooks that monitor task health, and enforced always-on OS-level sandboxing. If the sandbox is unavailable, the harness disables itself rather than running tools unprotected—a contrast with open-source harnesses that run commands with the user's full permissions by default.

On Perplexity's internal Local Knowledge Work Bench—53 tasks spanning deep research, financial analysis, and document creation—Computer running Qwen 3.8 27B on a DGX Spark scored 82.6%, versus 77.6% for the open-source Pi harness and 74.0% for Hermes running the identical model. Perplexity's post-trained PPLX 27B pushed the score to 85.4%. The gaps widen dramatically on harder tasks: on BrowseComp, a web research benchmark, Computer hit 66.7% accuracy versus 50.2% for Pi and 43.9% for Hermes, while using 51% less wall time and 70% fewer tokens than Pi. On multimodal document understanding, Computer scored 65.1% against Hermes' 34.6% and Pi's 13.9%.

The strategic logic behind the launch becomes clear when considering how AI workloads have changed. Chat was bursty—a question, an answer, done. Agents are different. Nader, Nvidia's director of developer technology, explained: "With agents, you want these agents always on if you can. You want the agents to really consume as many tokens as they can. What we're seeing is an insatiable demand for tokens, and that's something that makes local AI so great." An agent that runs for hours reviewing documents, verifying its own work, and iterating on analyses would rack up substantial API bills in the cloud. On a device the user already owns, the marginal cost of those tokens approaches zero. Perplexity's paper makes the enterprise version of this argument explicitly: as agents scale across individual workflows and entire organizations, token expenditure and data movement "become increasingly difficult to govern." Local-first execution addresses both at once—spend approaches zero because inference is free, and privacy is protected because sensitive tokens never leave the device boundary.

The hybrid middle ground offers perhaps the most commercially interesting result. On Terminal Bench 2.1, a challenging coding benchmark, the fully local Qwen model scored 59.6% at essentially zero marginal cost. Letting it escalate to a Claude Opus 5 "advisor" in the cloud raised the score to 73.0% at an estimated $0.415 per task. Running the frontier model alone scored 82.4% at $0.65 per task. Escalation, in other words, recovered roughly three-fifths of the gap to frontier performance at about two-thirds of the cost—and the user decides when that trade is worth making. Before any advisor call, the harness runs a PII classifier over the outgoing context and shows the user exactly what would leave the device. The remote model returns text guidance only; it never touches local files or tools.

The launch extends a partnership that has been building for more than a year. In June 2025, Nvidia and Perplexity announced a collaboration to bring sovereign AI models to European publishers and telecoms, part of CEO Jensen Huang's campaign to convince governments that every country needs a national intelligence infrastructure. The sovereign AI pitch—that data "belongs to your people, your country, your culture"—is philosophically the same argument Portable Computer makes at the scale of a single desk: intelligence you control, running on hardware you own. For Perplexity, which has raised capital at steadily escalating valuations while facing legal pressure from publishers over its content practices, including a lawsuit filed by The New York Times in December 2025, the product offers economics that don't depend on metering every token and a differentiated wedge into privacy-sensitive enterprises in law, healthcare, and finance. For Nvidia, it provides a killer app for DGX Spark, a device that, by the admission of attendees at Monday's briefing, has been easier to buy than to use. Questions remain: Perplexity's most impressive numbers come from its own internal benchmark, the 24GB VRAM floor excludes the vast majority of consumer PCs, and Apple silicon is conspicuously absent from the roadmap. But the direction of travel is unmistakable. Both companies are betting that advances in chips and open models will keep expanding what a box on a desk can do.