Google’s TurboQuant could cut LLM memory use sixfold, signaling a shift from brute-force scaling to efficiency and broader AI access.
Forbes