Google has introduced TurboQuant, a compression algorithm that reduces large language model (LLM) memory usage by at least 6x while boosting performance, targeting one of AI's most persistent ...