DeepSeek R1 Can Be Run at Home for $6,000, Generates 6 Tokens/Second, Faster Version Costs Over $100,000

A new computing setup allows users to run the DeepSeek R1 large language model (LLM) at home for approximately $6,000. This configuration, which does not require a GPU, includes a Gigabyte MZ73-LM0 or MZ73-LM1 motherboard, dual AMD EPYC 9004 or 9005 CPUs, and 768GB of DDR5 RAM. The system can generate over six token...