MMFlash: Faster Speculative Decoding for Vision-Language Models MMFlash, a speculative decoding method for vision-language models, is presented as faster than diffusion drafters such as DFlash, which propose an entire token block in a single forward pass but mainly focus on text-only tasks, leaving multimodal drafting underexplored. The method targets generation speed for multimodal models by pairing a lightweight drafter with parallel verification by a target model. Speculative decoding accelerates generation by letting a lightweight drafter propose tokens for parallel verification by a target model. Diffusion drafters such as DFlash propose an entire token block in a single forward pass, but mainly focus on text-only tasks, leaving multimodal drafting underexp