AI News

Perplexity Reportedly Prepares Hybrid Mode to Run AI Tasks Locally on Mac

Perplexity-Prepares-Hybrid-Mode-to-Run-AI-Tasks-Locally-on-Mac

Perplexity is working on a new Hybrid Mode for its Computer product on Mac, aiming to add local AI processing to a system that has mostly relied on cloud based models until now. With this update, Computer will keep managing and organizing tasks in the cloud, but it can send smaller, lighter subtasks to AI models that run directly on the user’s Mac. The idea is to combine the strength of cloud models with the privacy, savings, and local processing advantages that come with running smaller models directly on-device.

This update comes after Perplexity’s earlier announcement about hybrid agentic influence in June, and the company’s goal to expand beyond Windows to Mac and Linux. Perplexity also introduced Portable Computer for NVIDIA DGX Spark, which lets Computer run completely on local hardware. But for Mac, the approach is different. Instead of moving everything to the device, Hybrid Mode splits the work between local hardware and the cloud.

How Hybrid Mode will Work

Hybrid Mode lets Computer decide what parts of a bigger task are fit for local processing. Tougher tasks can keep using the cloud’s powerful models, while easier subtasks go to a model installed on the user’s Mac.

So, users won’t have to choose which model handles which part of a task, Computer handles all of that. It acts as the orchestration layer, running the whole process and sending parts to the local model automatically.

Local processing also reduces how much cloud computing a user may need for basic tasks. Perplexity says that any work done locally won’t use up Computer credits, and data can stay locally on the Mac instead of being required to upload over to the cloud.

Users can turn on Hybrid Mode in the Computer model selector. If one doesn’t have a local model set up, the system will help with the steps on how to install one.

Three downloadable models are in the works. The recommended Perplexity model is about 19 GB and needs at least 32 GB of memory. Qwen 32B is around 17.4 GB, with similar requirements. For a more lightweight option, there is a Gemma based model that’s 5.6 GB and works on Mac with 16 GB of memory. Perplexity has not revealed exactly what its own model is yet.

Privacy Gate Screens Data Before It Leaves the Mac

Alongside Hybrid Mode, Perplexity is preparing a feature called Privacy Gate. This tool checks information locally before anything gets sent to a cloud model.

Privacy Gate uses a separate local model to scan for personally identifiable information or other sensitive details. If it finds something sensitive, Computer can ask the user what to do with that information before sending it to the cloud.

This is important for people or companies worried about privacy laws like GDPR in the EEA. Instead of sending the complete task straight to the cloud, the system checks it first on the user’s computer.

A Shift Towards Local and Cloud AI

Perplexity’s plans show the Computer product shifting from a cloud-only AI agent to an orchestration system that can balance between cloud infrastructure and the user’s own hardware.

Perplexity’s Computer feature already works with local files, edits them, uses a computer, and browses using Perplexity’s Comet browser. The company also said Personal Computer is moving to Windows, with the broader hybrid technology aimed at all sorts of hardware.

The Mac version is not the same as the fully local setup with Portable Computer on NVIDIA DGX Spark. On the Mac, the cloud still handles the main orchestration, but certain tasks can now be moved locally.

The Mac controls for Hybrid Mode are not visible yet, and Perplexity has not announced a release date, it is also unclear if a specific subscription will be required. As for a fully local Mac orchestrator like Portable Computer on DGX Spark, there is no confirmation on that either.

You may also like to read – Perplexity Personal Computer on Mac: Features, Setup and How it Works

Devanshi Kashyap
Devanshi is a curious learner who enjoys exploring new ideas and expressing creativity through art.
You may also like
More in:AI News