inference optimization

DeepSeek's new AI model reduces prediction costs by 75%, cutting expenses from $1.68 to $0.42 per million tokens to enhance…