Perplexity has introduced a 'Hybrid Computing' feature in its Mac app, which can split the same task and process it collaboratively through cloud-based cutting-edge models and local large language models. Sensitive information is automatically allocated to local processing to protect privacy. This feature integrates a privacy classifier, allowing users to review isolated content. Locally, it offers Gemma E4B and two versions of the Qwen 3.6 model, supporting one-click installation. Tokens generated by local models are not charged additionally, and users can also add tasks to the queue via iPhone. While cloud models generate high-quality results, they also come with higher costs. This feature provides an adjustable solution, with the final decision-making power in the hands of users. Currently, this feature is only supported on Macs equipped with Apple Silicon chips, running macOS 15, and with at least 32GB of unified memory. It is available to Pro and Max subscribers as well as enterprise customers.