Tag
The author calculates the token cost and break-even period of running large models on a Mac Studio, concluding that it is not cost-effective for ordinary users to buy a Mac for personal large model use, and suggests that using APIs or renting GPUs is more economical.