Both models trade word-by-word generation for parallel denoising. Only one of them does it without losing intelligence in the ...
PALO ALTO, Calif., February 24, 2026--(BUSINESS WIRE)--Inception, the company behind the first commercial diffusion large language models (dLLMs), today announced the launch of Mercury 2, the fastest ...