Perplexity has recently launched Portable Computer, a local-first version of its Perplexity Computer that is designed to run directly on users’ local machines. The new tool aims at users who need to perform complex AI workflows while keeping sensitive code, files, and private information on their local device.
Developed in partnership with NVIDIA, Portable Computer runs on the NVIDIA DGX Spark ™, NVIDIA’s desktop-sized AI computer, with Linux support at Launch. NVIDIA RTX GPU PC support and Windows support are both planned to follow.
How Does Portable Computer Handle Local-first and Cloud AI?
Portable Computer is designed as a complete AI agent. It combines a local AI model with an inference engine, an orchestrator, a tool router, a scheduler, a task queue, a search index, application connectors, and an isolated sandbox.
This allows it to analyze files, search documents and code, and manage multi-step workflows directly on the device. Portable Computer runs on the NVIDIA DGX Spark with Qwen 3.8 27B and Perplexity’s post-trained PPLX 27B model, a post-trained version of the Qwen model.
NVIDIA Nemotron 3.5 Lightning, a 30-billion-parameter open model, is expected to be added to the model picker soon. Perplexity also plans to expand support for models and inference servers.
Perplexity says that Portable Computer’s local orchestrator can escalate a task to the cloud for current information, browser access, connected applications, or one of 15+ frontier models for advanced reasoning.
A Portable Computer can connect to services like Google Drive, Slack, Gmail, and GitHub. Users can analyze local files, review GitHub issues, or process information from connected applications while maintaining control over when data is shared externally.
Whenever a task needs to send content from the device to a cloud service, Portable Computer asks the user for permission before moving ahead.
One noteworthy feature is that when a task needs to be done outside the local machine, a Portable Computer can run a Perplexity search or deep research. Users can access cited web findings while keeping everything running on the local machine.
Portable Computer: A Major Focus on On-Device and Privacy
The system also includes a dictation model that runs locally with the NVIDIA Nemotron 3.5 ASR Model. Perplexity says the audio transcriptions and actions on the file remain on the machine, helping keep sensitive information and conversations private. A user can speak through a client situation or dictate over a private codebase without the audio touching the cloud.
Security is another key focus of Perplexity Computer. Code and tool execution operate inside isolated sandbox environments with controlled access to files and connected applications. Perplexity also says the system blocks unauthorized network connections.
Perplexity’s Move Towards Local AI
According to Perplexity AI, “As models get stronger and chips get faster, more people will run complex workflows on their own machines.”
Portable Computer is currently available to Pro and Max Subscribers on the NVIDIA DGX Spark. The first release will be available on Linux. Windows support is coming soon.
Portable Computer marks Perplexity’s significant push towards local-first AI. It combines on-device privacy and cost control with optional cloud-based intelligence for users who require powerful reasoning or up-to-date information.
To stay updated on the latest news in the tech world, don't miss out on our website.
Recommended For You:
BigCommerce and Feedonomics Team Up with Perplexity to Help Brands Excel at AI Product Search
OpenAI and Perplexity Debut in AI Shopping Assistants for Smarter and Tailored Purchase Experience
