“Modern, efficient large language models can now be run locally on a laptop without requiring data center compute.”