Perplexity Mac Hybrid mode: Local models handle subtasks

testingcatalog · x · 2026-08-31

Perplexity is launching a Hybrid mode for Mac, allowing local models to handle specific subtasks to save costs. Supported models include Gemma 4 (16GB RAM), Qwen 2.5 32B (32GB RAM), and a proprietary Perplexity model (32GB RAM). Cloud orchestrates heavy reasoning while local inference handles lighter tasks. A new 'Privacy Gate' feature runs locally to inspect data for PII before cloud transmission.

Related event: Perplexity Brings Hybrid Mode to Mac With Local Models(2 posts)→

Original post →

More from Infra

Infra channel →