UC San Diego researchers have developed DFlash, a block diffusion model that drafts whole token blocks in a single pass ...

UC San Diego researchers have developed DFlash, a block diffusion model that drafts whole token blocks in a single pass for speculative decoding. The technique delivers up to 15x higher throughput on NVIDIA Blackwell GPUs compared to traditional autoregressive methods. It works by having a small draft model propose token blocks that the larger target model verifies in parallel, making AI inference significantly faster for coding agents and reasoning models. https://www.marktechpost.com/2026/06/24/dflash-speculative-decoding-drafts-whole-token-blocks-in-parallel-for-up-to-15x-higher-throughput-on-nvidia-blackwell/ #AIagent #AI #GenAI #AIInfrastructure