Google's TurboQuant AI-compression algorithm can reduce LLM memory usage by 6x

(arstechnica.com) by gmays | Mar 27, 2026 | 0 comments on HN