Both models trade word-by-word generation for parallel denoising. Only one of them does it without losing intelligence in the ...

Google's open-source diffusion language model generates 256 tokens in parallel and self-corrects, hitting 4x speed on one GPU at a cost to quality.

Add Decrypt as your preferred source to see more of our stories on Google. Google released DiffusionGemma, a free open-weight model that generates entire 256-token blocks simultaneously via text ...

“By combining our advanced Mercury model development with Civo’s UK sovereign cloud infrastructure, we’re creating a trusted, homegrown AI ecosystem that meets the highest standards of security, ...

PALO ALTO, Calif.--(BUSINESS WIRE)--Inception, the company behind the first commercial diffusion large language models (dLLMs), today announced the launch of Mercury 2, the fastest reasoning LLM and ...

AI startup Inception Labs launches the first diffusion-based reasoning AI model. It's fast and cheap. Unlike conventional language models, Mercury 2 refines multiple text blocks simultaneously instead ...

Awesome Diffusion Language Models One of the most starred, comprehensive and up-to-date collections of Diffusion Language Model papers, code and resources! If you find this repository helpful, please ...

Diffusion models gradually refine and produce a requested output, sometimes starting from random noise—values generated by the model itself—and sometimes working from user-provided data. Think of ...

What if the future of text generation wasn’t just faster, but smarter and more adaptable? Enter Gemini Diffusion, a new approach that challenges the long-standing dominance of autoregressive models.