Perplexity Portable Computer Arrives on Windows, Runs AI Locally on RTX GPUs

Summary

Perplexity has expanded its Portable Computer, a local AI agent that runs entirely on-device, to Windows PCs equipped with NVIDIA RTX GPUs with at least 24GB of VRAM. The tool handles multi-step tasks using local models, keeping sensitive data off the cloud and completing work without consuming Perplexity Computer credits. It installs through the Microsoft Store and is available to Perplexity Pro, Max, and Enterprise subscribers. The Windows launch follows a debut on Linux and NVIDIA DGX Spark hardware three weeks prior.

Why it matters

On-device AI agents with no cloud dependency address two concerns that keep some users from going all-in on AI tools: data privacy and unpredictable credit costs. The 24GB VRAM requirement limits this to high-end RTX cards (RTX 4090, RTX 5090, and RTX PRO workstation hardware), but for users with that setup, Perplexity’s agent now runs without sending anything to a remote server.

HTD Says

Perplexity’s local AI agent now runs on Windows, and if you have a high-end RTX machine and have been sitting out the AI agent wave because of privacy concerns, this version is for you. I have been saying that local LLMs and local AI are the direction that AI seems to be heading. While not as powerful or as advanced as public, subscription-based AI, it’s a viable alternative for many consumer tasks (and plenty of business operations) to usage- and token-based consumption.

It is smart of NVIDIA to co-market and push this Windows solution along. Got an NVIDIA GPU? You are golden. If not, it’s time to upgrade, just so that you can do AI better locally.

The Perplexity Windows solution is limited to higher-end cards, but I view this as a test of the waters. I expect more local AI solutions to hit the market, and I wouldn’t count out the Mac Mini fanatics, as that new hardware just came out as well.

Source:

NVIDIA Blog