ToolsAugust 25, 2026via MarkTechPost
Perplexity Ships Portable Computer on NVIDIA DGX Spark: Local Harness, OS-Enforced Sandbox, and Zero Per-Token Cost for Local Steps
Why it matters
A major AI product company is moving inference workloads from cloud APIs to on-premises hardware, signaling a shift in how practitioners deploy and cost AI — and directly competing with hosted API models.
Key signals
- Perplexity releases 'Portable Computer' product
- Runs on NVIDIA DGX Spark hardware
- Packages local models, harness, OS-enforced sandbox, connectors
- Zero per-token cost for local inference steps
- On-premises deployment model vs. cloud APIs
- Published August 25, 2026
The hook
Perplexity ships a local-first AI compute box: models, sandbox, and zero per-token cost bundled on DGX Spark hardware.
Perplexity releases Portable Computer, packaging local models, harness, sandbox, and connectors into one system running on NVIDIA DGX Spark.
The post Perplexity Ships Portable Computer on NVIDIA DGX Spark: Local Harness, OS-Enforced Sandbox, and Zero Per-Token Cost for Local Steps appeared first on …