DFlash in llama.cpp: 4.44x Faster Local Inference on Qwen 3.6 27B
DFlash, merged into llama.cpp, uses block-diffusion drafting to boost Qwen 3.6 27B inference speed, hitting 4.44x at 36K context with near-lossless quality.
llama.cppspeculative-decodinglocal-aiqwen