Perplexity: Hybrid Local-Server Inference Orchestrator Arrives on Perplexity Computer in July

What's New

• Hybrid orchestrator: Automatic routing of AI tasks between local device and cloud
• Sensitive data on-device: Financial records, health files stay on device
• Compute-heavy tasks in cloud: Frontier models for complex tasks
• Model-agnostic and chip-agnostic: Runs on Intel Core Ultra Series 3 and NVIDIA RTX Spark
• No manual configuration: Compact local model acts as router\n\n## Why It Matters

This solves the fundamental tension between privacy and performance in AI. Sensitive data never leaves the device, while complex tasks still benefit from frontier models in the cloud. It's a step closer to 'local AI' for everyday users.\n\n## How to Try It
Feature arrives in Perplexity Computer in July 2026, initially on Windows. Personal Computer is already on Mac, Windows waitlist open at perplexity.ai.