Industry

New Dflash drafting model for the 27b Lets gooooo https://huggingface.co/z-lab/Qwen3.6-27B-DFlash

A new Dflash drafting model based on Qwen 3.6 with 27 billion parameters has been released on Hugging Face, available at z-lab/Qwen3.6-27B-DFlash. This model likely implements speculative decoding or

DGX agentx-post
industryclem-delangue--x
Loading related sources…