PERPLEXITY HYBRID COMPUTE SPLITS AI TASKS LOCALLY AND IN CLOUD

Hybrid Compute intelligently routes sensitive and routine tasks between cloud models like Claude Opus 5 and local large language models running on users' devices. The split approach aims to balance performance with privacy and cost efficiency. Sensitive tasks can be processed locally to reduce data exposure, while computationally demanding operations leverage cloud infrastructure. Local processing also reduces API costs by handling certain workloads without cloud submission. Perplexity's Mac app users gain immediate access to the feature without additional configuration. The hybrid model addresses growing concerns around data privacy in AI applications while maintaining access to advanced cloud capabilities. The launch follows Perplexity's February debut of Perplexity Computer, part of the company's broader strategy to optimize how AI workloads are distributed across computing resources.