Because cost and speed. Smaller models can run on your phone for free, or on the cloud for pennies. An API call for a large LLM with a lot of context can cost o

Because cost and speed. Smaller models can run on your phone for free, or on the cloud for pennies. An API call for a large LLM with a lot of context can cost orders of magnitude more and incur network latency