---
format: "aidr-story-markdown/v1"
id: "7043cdd25b6ec0e7037c00cbdf122a27554da76a4accef09fb28779429b0c489"
canonical_url: "https://aidr.today/7043cdd2?lang=en"
title: "Perplexity Ships First Local AI Agent Runtime for Nvidia DGX Spark to Hit 73% Bench Score"
lang: "en"
requested_lang: "en"
available_langs: ["en","vi"]
translation_fallback: null
fallback_fields: []
published_at: "2026-08-25T16:48:16.000Z"
category: "Products"
topics: ["perplexity","nvidia","agent","inference","qwen"]
source_urls: ["https://huggingnews.com/ai/perplexity-ships-first-local-ai-agent-runtime-for-nvidia-dgx-spark-to-hi-dfff715c","https://x.com/perplexity_ai/status/2092268362386780270","https://x.com/perplexity_ai/status/2092268381449904327","https://x.com/perplexity_ai/status/2092268398319481039","https://x.com/perplexity_ai/status/2092268411074310485","https://x.com/nvidia/status/2092269109086126575","https://x.com/AravSrinivas/status/2092270041471598820","https://x.com/AravSrinivas/status/2092271842816053390"]
summary: "Portable Computer is a new local AI agent stack that executes the entire runtime—including the orchestrator LLM, subagent LLM, and agent harness—on-device using Nvidia DGX Spark. Available to Perplexity Pro and Max subscribers, the system initially supports the post-trained PPLX 27B model and Qwen 3.8 27B. The platform offers a one-click local inference setup to eliminate cloud dependencies for standard agentic workflows. The system utilizes a hybrid inference model that routes complex reasoning tasks to cloud frontier models after manual user approval and PII filtering. This escalation increases the system's Terminal Bench 2.1 score from 59.6% to 73.0% at a cost of $0.415 per rollout. Support for Nvidia Nemotron 3.5 Lightning is planned, and remote models are restricted to returning text guidance without direct access to local files."
---

# Perplexity Ships First Local AI Agent Runtime for Nvidia DGX Spark to Hit 73% Bench Score

> [Open the canonical story](<https://aidr.today/7043cdd2?lang=en>)

**Published:** 2026-08-25T16:48:16.000Z
**Category:** Products
**Topics:** perplexity, nvidia, agent, inference, qwen

## Summary

Portable Computer is a new local AI agent stack that executes the entire runtime—including the orchestrator LLM, subagent LLM, and agent harness—on\-device using Nvidia DGX Spark\. Available to Perplexity Pro and Max subscribers, the system initially supports the post\-trained PPLX 27B model and Qwen 3\.8 27B\. The platform offers a one\-click local inference setup to eliminate cloud dependencies for standard agentic workflows\. The system utilizes a hybrid inference model that routes complex reasoning tasks to cloud frontier models after manual user approval and PII filtering\. This escalation increases the system's Terminal Bench 2\.1 score from 59\.6% to 73\.0% at a cost of $0\.415 per rollout\. Support for Nvidia Nemotron 3\.5 Lightning is planned, and remote models are restricted to returning text guidance without direct access to local files\.

## Sources

- [Story source](<https://huggingnews.com/ai/perplexity-ships-first-local-ai-agent-runtime-for-nvidia-dgx-spark-to-hi-dfff715c>)
- [Story source](<https://x.com/perplexity_ai/status/2092268362386780270>)
- [Supporting source](<https://x.com/perplexity_ai/status/2092268381449904327>)
- [Supporting source](<https://x.com/perplexity_ai/status/2092268398319481039>)
- [Supporting source](<https://x.com/perplexity_ai/status/2092268411074310485>)
- [Story source](<https://x.com/nvidia/status/2092269109086126575>)
- [Supporting source](<https://x.com/AravSrinivas/status/2092270041471598820>)
- [Supporting source](<https://x.com/AravSrinivas/status/2092271842816053390>)