NVIDIA Developer

DGX Spark Live: Perplexity Portable Computer Goes Local

Published 2026-08-31 · Duration 26:37

Summary

Perplexity introduced Portable Computer, an agent platform designed to run fully on-device using NVIDIA DGX Spark. This system brings complex, multi-step AI workflows—including inference and agent harnessing—to a simple local interface, prioritizing data privacy by keeping sensitive processing offline. While defaulting to local models (like Qwen 27B), the architecture supports controlled escalation to frontier cloud models only when necessary or permitted.

Download summary

Key takeaways

  1. Local-First AI Architecture 3:42

    Portable Computer runs the entire stack—including agent harness and inference—locally on DGX Spark, eliminating token caps and metered compute for local tasks. This ensures sensitive data (e.g., tax documents) remains fully private [0:03:42].

  2. Simplified Agent Experience 1:58

    The platform abstracts away the underlying complexity of agent harnesses and inference, providing users with a simple interface to execute sophisticated AI tasks without needing deep knowledge of the stack [0:01:58].

  3. Hybrid Scalability 2:36

    While designed for local operation, Portable Computer is 'local first' but can escalate to use frontier cloud models (e.g., Anthropic, OpenAI) and connect via a robust connector ecosystem when required [0:02:37].

Technical details

  • Hardware & Compatibility 238s

    The initial launch targets the NVIDIA DGX Spark. However, Perplexity is working to expand compatibility, expecting support for RTX devices with at least 24 GB of VRAM [0:03:58].

  • Model Stack and Performance 108s

    The system utilizes post-trained models like the Perplexity 27 billion parameter Qwen model (for initial demos) and Nemotron 3.5 Lightning. The architecture is designed to manage inference locally while maintaining frontier-level utility [0:01:48, 0:03:16].

  • Agent Workflow Management 163s

    The system features a built-in mechanism for controlled cloud escalation, requiring user approval before sending traffic to external/frontier models. It also includes advanced sandboxing for safe execution [0:02:45].

Mentioned resources

  • DGX Spark (Hardware Platform)
  • Perplexity Pro/Max Subscription (Access Requirement)
  • NVIDIA Developer Website (Information Source)

Channel & topics

Watch on YouTube · Back to latest

This independent, AI-assisted summary is provided for commentary and informational purposes. It may contain errors or omit important context. Please watch the original video for the creator's complete presentation. Video, thumbnail, and related copyrights belong to their respective owners.