> ## Content Index
> Fetch the complete content index at: https://www.testingcatalog.com/llms.txt
> Use this file to discover other available public pages before exploring further.

# Perplexity prepares Hybrid mode for Computer on Mac
- URL: https://www.testingcatalog.com/perplexity-prepares-hybrid-mode-for-computer-on-mac/
- Published: 2026-08-31T08:48:58.000Z
- Updated: 2026-08-31T08:48:58.000Z
- Description: What we know so far: Hybrid mode will delegate some subtasks to models hosted locally in order to save costs. Privacy Gate will evaluate data for PII before sending it to the cloud.
- Author: Alexey Shabanov
- Tags: Perplexity News, Latest AI News, AI Rumours

Perplexity is preparing to bring local inference to a much wider group of Mac users through a new Hybrid Mode for Perplexity Computer. The company previously announced hybrid agentic inference in June and said support would extend beyond Windows to Macs and Linux. More recently, it launched Portable Computer for NVIDIA DGX Spark, where Computer can run entirely on local hardware.

> Today we’re launching Portable Computer on [@NVIDIA](https://x.com/nvidia?ref%5Fsrc=twsrc%5Etfw&ref=testingcatalog.com) DGX Spark.  
>  
> Portable Computer is a fully local version of Perplexity Computer, where the entire runtime: orchestrator LLM, subagent LLM, agent harness all run on your local hardware. No cloud dependency. [pic.twitter.com/plVWz5PaAw](https://t.co/plVWz5PaAw?ref=testingcatalog.com)
> 
> — Perplexity (@perplexity\_ai) [August 25, 2026](https://x.com/perplexity%5Fai/status/2092268362386780270?ref%5Fsrc=twsrc%5Etfw&ref=testingcatalog.com)

On Mac, the implementation appears different. Hybrid Mode lets Computer continue orchestrating in the cloud while automatically delegating suitable subtasks to a model running locally. Cloud models handle heavier reasoning, while local inference handles lighter work without consuming Computer credits and keeps the associated data on the Mac.

![Perplexity](https://storage.ghost.io/c/2a/1b/2a1b1782-8506-4d7d-bf53-ad3fb52e2a0f/content/images/2026/08/Screenshot-2026-08-30-at-01.00.39.png)

![Perplexity](https://storage.ghost.io/c/2a/1b/2a1b1782-8506-4d7d-bf53-ad3fb52e2a0f/content/images/2026/08/Screenshot-2026-08-30-at-00.59.13.png)

Users can enable Hybrid Mode from the Computer model selector and configure local inference if no model is installed. Three downloadable options are currently being prepared: a recommended Perplexity model around 19 GB requiring at least 32 GB of memory, a roughly 17.4 GB Qwen 32B option with similar requirements, and a smaller Gemma-based model around 5.6 GB that can run on 16 GB Macs. The exact identity of Perplexity's own model remains unclear.

![Perplexity](https://storage.ghost.io/c/2a/1b/2a1b1782-8506-4d7d-bf53-ad3fb52e2a0f/content/images/2026/08/Screenshot-2026-08-31-at-00.45.36.png)

Another feature, Privacy Gate, adds a separate local model that inspects data before it is sent to the cloud. If it detects personal or sensitive information, Computer can ask the user how to handle it before transmission. This could be particularly relevant to users and companies operating under GDPR requirements in the EEA.

![Perplexity](https://storage.ghost.io/c/2a/1b/2a1b1782-8506-4d7d-bf53-ad3fb52e2a0f/content/images/2026/08/Screenshot-2026-08-31-at-00.45.05.png)

Computer is evolving from a cloud-only agent into an orchestration layer spanning cloud infrastructure and user-controlled hardware. The company has already publicly described this architecture as hybrid agentic inference, positioning local processing around privacy and lower inference costs while retaining access to frontier models.

The Mac controls described here remain hidden, and [Perplexity](https://www.testingcatalog.com/tag/perplexity/) has not announced their release date or whether access will be limited by subscription tier. A fully local Mac orchestrator, similar to Portable Computer on DGX Spark, also does not appear to be confirmed yet.