Perplexity Launches Local AI Agent 'Portable Computer' with Nvidia

Perplexity has unveiled Portable Computer, a new version of its agentic platform that relies on local AI models to handle requests. [1] Developed in close partnership with Nvidia, the system runs entirely on user-owned hardware, such as the DGX Spark desktop supercomputer and Linux machines equipped with RTX GPUs. [3] Debuting on Tuesday, the platform offers access to the standard AI commands and tools through which Perplexity's agent can carry out tasks with little or no intervention. [1]

Perplexity packages the local model, inference engine, agent harness, tool sandbox, and app connectors as one packaged system. [5] This packaging removes the usual work of standing up an inference server and wiring tools by hand. [5] Nate, Perplexity's vice president of engineering for infrastructure and enterprise, stated that the company brought the exact same UI to a fully local app. [3] Users are able to select either the Qwen 3.8 27B or PPLX 27B models to run locally on their machines. [5]

The launch represents one of the most aggressive attempts yet to move serious AI agent workloads off the cloud and onto local devices. [3] Because the model, the user's files, and the work itself can all stay on the machine, work completed locally consumes no billing credits. [3] Every task begins on the device by default, and the orchestrator stops and asks before sending a single step to one of 15 or more cloud models. [5] If needed, users can still connect to external services like Google Drive, Gmail, Slack, and GitHub to incorporate their cloud-based files and information. [1]

To deploy this shipping software, users must meet a strict hardware gate requiring a GB10-class box or an RTX GPU with 24 GB of VRAM. [5] Local AI models can run faster and respond to your requests more quickly than online models. [1] Nvidia's director of developer technology, Nader, noted that local AI has transitioned from hobbyists running quantized models to practical applications powered by new open source models. [3]

The introduction of Portable Computer provides a deployable solution for industries like finance, legal, and healthcare where data residency or contractual confidentiality blocks cloud inference. [5] By partnering with Nvidia, Perplexity is leveraging the hardware maker's push to sell systems beyond AI data centers directly to enterprises and well-funded startups. [3] This shift to local execution eliminates per-token charges for local steps, altering the economics of continuous agent workflows. [5] It also provides faster performance compared to standard cloud-based models while maintaining a familiar interface. [1][3] What the sources don't address: Whether Perplexity intends to optimize Portable Computer for lower-tier consumer hardware outside of high-end NVIDIA configurations.