AI & Enterprise

Perplexity adds \'hybrid compute\' to Mac app, splits work across cloud and local AI

Perplexity has introduced \'hybrid compute\' in its Mac app, a feature that keeps sensitive information on a Mac device while distributing other work to cloud AI. It splits a task between cloud-based models and locally run large language models, and can automatically check whether files contain sensitive content before sending data. The company said cloud-only performance is generally better, but users can choose based on data protection and cost.

AI & Enterprise

Agent era accelerates multi-model use; small models set to take larger share

Multi-model use inside generative AI services is already common, with different-sized models dividing work, a Ravlub researcher said. Small models handle routine tasks such as search, information gathering and summarisation, while frontier models take key decisions and generate outputs. As agents spread, repeatedly calling top-performance models can push inference costs higher. He said small models will not replace frontier models, citing scaling, but role-based use of multiple model sizes will expand.