Perplexity may have a way to run LLMs on consumer hardware: How it works

Perplexity has launched a hybrid compute feature for its Mac app. It lets a 35-billion-parameter model run partly on your device, not only in the cloud. The company is also open-sourcing Lily, its local inference engine for Apple Silicon. For everyday users, this means big AI models can work on the computer you already own. Try the feature to see if it suits your needs.