Google Research publishes TurboQuant paper claiming 80% AI cost reduction and 6x memory reduction

Novel compression algorithm unveiled that reduces LLM memory footprint by 6x

Comprehensive review published categorizing five optimization techniques for million-token LLM inference