MiniMax-M2.5-DFlash-GGUF

MiniMax M2.5 DFlash GGUF

GGUF quantizations of z-lab DFlash draft model for MiniMax M2.5.

Use with BeeLlama.cpp, a llama.cpp fork with advanced quantization features.

---

MiniMax-M2.5-DFlash

Paper | GitHub | Blog

DFlash is a speculative decoding method that uses a lightweight block diffusion model to draft multiple tokens in parallel. This is the drafter model, which must be paired with MiniMaxAI/MiniMax-M2.5.

<div align="center"> <img src="assets/dflash_system.png" alt="DFlash Architecture" width="85%"> </div>

Quick Start

Installation

vLLM:

Check out vLLM issue #46105.

SGLang:

uv pip install "git+https://github.com/sgl-project/sglang.git#subdirectory=python"

Launch Server

vLLM:

Check out vLLM issue #46105.

SGLang:

python -m sglang.launch_server \
  --model-path MiniMaxAI/MiniMax-M2.5 \
  --tp-size 4 \
  --speculative-algorithm DFLASH \
  --speculative-draft-model-path z-lab/MiniMax-M2.5-DFlash \
  --attention-backend trtllm_mha \
  --speculative-draft-attention-backend fa4 \
  --mem-fraction-static 0.8 \
  --trust-remote-code \
  --host 0.0.0.0 \
  --port 30000

Usage

For SGLang, use port 30000.

from openai import OpenAI

client = OpenAI(base_url="http://localhost:30000/v1", api_key="EMPTY")

response = client.chat.completions.create(
    model="MiniMaxAI/MiniMax-M2.5",
    messages=[{"role": "user", "content": "Write a quicksort in Python."}],
    max_tokens=4096,
    temperature=0.0,
    extra_body={"chat_template_kwargs": {"enable_thinking": True}},
)
print(response.choices[0].message.content)

Benchmark Results

Setup: 4 NVIDIA B200 GPUs per server/run, SGLang, tensor parallel size 4, target attention backend trtllm_mha, draft attention backend fa4, thinking enabled, max output length 4096, greedy decoding. Concurrency 1 uses 128 prompts; concurrency 32 uses 1024 prompts.

Throughput

Generated tokens/sec

Block Size = 8

Acceptance Length

Acknowledgements

Special thanks to David Wang for his outstanding engineering support on this project. We are also grateful to Modal, InnoMatrix, and Yotta Labs for providing the compute resources used to train this draft model.

Citation

If you find DFlash useful, please cite our work. To share feedback on DFlash or request new model support, please fill out this form: DFlash Feedback.

@article{chen2026dflash,
  title   = {{DFlash: Block Diffusion for Flash Speculative Decoding}},
  author  = {Chen, Jian and Liang, Yesheng and Liu, Zhijian},
  journal = {arXiv preprint arXiv:2602.06036},
  year    = {2026}
}