Qwen3.8-Flash-Next-DFlash2-GGUF

Blog | GitHub

This repository contains GGUF conversions of incoai/Qwen3.8-Flash-Next-DFlash2, the DFlash 2 draft model for Qwen/Qwen3.8-Flash-Next. It is not a standalone language model: it runs inside a speculative decoding server and drafts tokens for the target model to verify.

DFlash 2 is a block-diffusion drafter for speculative decoding. It predicts a whole block of tokens in a single pass and keeps the top candidates at every position. A lightweight selector then traces one coherent path through them. Two-tap dynamic convolutions in the backbone keep the draft from decaying toward the end of the block. Decoding is lossless: greedy output matches the target model exactly, and sampling preserves its distribution.

Quick Start

Download a quantization:

hf download incoai/Qwen3.8-Flash-Next-DFlash2-GGUF \
  Qwen3.8-Flash-Next-DFlash2-Q4_K_M.gguf \
  --local-dir ./Qwen3.8-Flash-Next-DFlash2-GGUF

llama.cpp's DFlash 2 support (PR #27342) cannot drive this draft yet. Qwen3.8-Flash-Next carries its hidden state as four hyper-connection streams, and the draft is trained on their mean, which llama.cpp does not compute for the draft yet. Until it does, serve the safetensors checkpoint with SGLang.

See the blog post for other engines and more details.

Evaluation

Acceptance lengths for these GGUFs will be added once llama.cpp supports this target. Full evaluations of the base checkpoint are on the main model card.

Citation

If you find DFlash 2 useful, please cite:

@misc{inco2026dflash2,
  title  = {{DFlash 2: Keep Drafting Parallel}},
  author = {{Inco AI}},
  year   = {2026},
  month  = {August},
  url    = {https://inco.ai/blog/dflash2/}
}

Please also cite the original DFlash paper:

@inproceedings{chen2026dflash,
  title     = {{DFlash: Block Diffusion for Flash Speculative Decoding}},
  author    = {Chen, Jian and Liang, Yesheng and Liu, Zhijian},
  booktitle = {International Conference on Machine Learning (ICML)},
  year      = {2026}
}