Instructions to use Akicou/Qwen3.8-27B-DFlash2-GGUF with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Notebooks
Google Colab
Kaggle
Local Apps Settings

How to use Akicou/Qwen3.8-27B-DFlash2-GGUF with llama.cpp:

Install (macOS, Linux)

curl -LsSf https://llama.app/install.sh | sh
# Start a local OpenAI-compatible server with a web UI:
llama serve -hf Akicou/Qwen3.8-27B-DFlash2-GGUF:Q4_K_M
# Run inference directly in the terminal:
llama cli -hf Akicou/Qwen3.8-27B-DFlash2-GGUF:Q4_K_M

Install from WinGet (Windows)

winget install llama.cpp
# Start a local OpenAI-compatible server with a web UI:
llama serve -hf Akicou/Qwen3.8-27B-DFlash2-GGUF:Q4_K_M
# Run inference directly in the terminal:
llama cli -hf Akicou/Qwen3.8-27B-DFlash2-GGUF:Q4_K_M

Build from source code

git clone https://github.com/ggml-org/llama.cpp.git
cd llama.cpp
cmake -B build
cmake --build build -j --target llama-app
# Start a local OpenAI-compatible server with a web UI:
./build/bin/llama serve -hf Akicou/Qwen3.8-27B-DFlash2-GGUF:Q4_K_M
# Run inference directly in the terminal:
./build/bin/llama cli -hf Akicou/Qwen3.8-27B-DFlash2-GGUF:Q4_K_M

Use Docker

docker model run hf.co/Akicou/Qwen3.8-27B-DFlash2-GGUF:Q4_K_M

Llama new
LM Studio
Jan

vLLM

How to use Akicou/Qwen3.8-27B-DFlash2-GGUF with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "Akicou/Qwen3.8-27B-DFlash2-GGUF"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "Akicou/Qwen3.8-27B-DFlash2-GGUF",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/Akicou/Qwen3.8-27B-DFlash2-GGUF:Q4_K_M

Ollama
How to use Akicou/Qwen3.8-27B-DFlash2-GGUF with Ollama:
```
ollama run hf.co/Akicou/Qwen3.8-27B-DFlash2-GGUF:Q4_K_M
```
Unsloth Desktop

How to use Akicou/Qwen3.8-27B-DFlash2-GGUF with Pi:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf Akicou/Qwen3.8-27B-DFlash2-GGUF:Q4_K_M

Configure the model in Pi

# Install Pi:
npm install -g @earendil-works/pi-coding-agent
# Add to ~/.pi/agent/models.json:
{
  "providers": {
    "llama-cpp": {
      "baseUrl": "http://localhost:8080/v1",
      "api": "openai-completions",
      "apiKey": "none",
      "models": [
        {
          "id": "Akicou/Qwen3.8-27B-DFlash2-GGUF:Q4_K_M"
        }
      ]
    }
  }
}

Run Pi

# Start Pi in your project directory:
pi

Docker Model Runner
How to use Akicou/Qwen3.8-27B-DFlash2-GGUF with Docker Model Runner:
```
docker model run hf.co/Akicou/Qwen3.8-27B-DFlash2-GGUF:Q4_K_M
```

Lemonade

How to use Akicou/Qwen3.8-27B-DFlash2-GGUF with Lemonade:

Pull the model

# Download Lemonade from https://lemonade-server.ai/
lemonade pull Akicou/Qwen3.8-27B-DFlash2-GGUF:Q4_K_M

Run and chat with the model

lemonade run user.Qwen3.8-27B-DFlash2-GGUF-Q4_K_M

List all available models

lemonade list

Hermes Agent

How to use Akicou/Qwen3.8-27B-DFlash2-GGUF with Hermes Agent:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf Akicou/Qwen3.8-27B-DFlash2-GGUF:Q4_K_M

Configure Hermes

# Install Hermes:
curl -fsSL https://hermes-agent.nousresearch.com/install.sh | bash
hermes setup
# Point Hermes at the local server:
hermes config set model.provider custom
hermes config set model.base_url http://127.0.0.1:8080/v1
hermes config set model.default Akicou/Qwen3.8-27B-DFlash2-GGUF:Q4_K_M

Run Hermes

hermes

Atomic Chat

OpenClaw

How to use Akicou/Qwen3.8-27B-DFlash2-GGUF with OpenClaw:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf Akicou/Qwen3.8-27B-DFlash2-GGUF:Q4_K_M

Configure OpenClaw

# Install OpenClaw:
npm install -g openclaw@latest
# Register the local server and set it as the default model:
openclaw onboard --non-interactive --mode local \
  --auth-choice custom-api-key \
  --custom-base-url http://127.0.0.1:8080/v1 \
  --custom-model-id "Akicou/Qwen3.8-27B-DFlash2-GGUF:Q4_K_M" \
  --custom-provider-id llama-cpp \
  --custom-compatibility openai \
  --custom-text-input \
  --accept-risk \
  --skip-health

Run OpenClaw

openclaw agent --local --agent main --message "Hello from Hugging Face"

Qwen3.8-27B-DFlash2 GGUF

GGUF conversions of z-lab/Qwen3.8-27B-DFlash2, the DFlash 2 draft model for Qwen/Qwen3.8-27B.

This is not a standalone language model. It is a draft model for speculative decoding: it drafts blocks of tokens that Qwen3.8-27B then verifies. Loading it on its own fails with dflash requires ctx_other to be set, which is the expected response for a draft model without its target. Run it next to the target model with --spec-type draft-dflash.

DFlash 2 predicts a whole block of tokens in one pass and keeps the top candidates at every position. A selector then traces one path through the candidates. Two-tap dynamic convolutions in the backbone keep draft quality from dropping toward the end of the block. Decoding is lossless: greedy output matches the target model, and sampling preserves its distribution. The z-lab repository is a mirror of incoai/Qwen3.8-27B-DFlash2. The DFlash 2 paper and reference code are at inco.ai/blog/dflash2 and github.com/z-lab/dflash.

Files

File	Quant	Size	Bytes	BPW	SHA-256
`Qwen3.8-27B-DFlash2-BF16.gguf`	BF16	3.60 GiB	3,860,292,992	16.00	`30c30ad2def19d4834dd8e2fe5505c2d4b954bafe2b43f7a0e502d5d9fb9048b`
`Qwen3.8-27B-DFlash2-Q6_K.gguf`	Q6_K	1.48 GiB	1,590,412,672	6.57	`59c09bfdd28aa5dd3dfe18c8a899d008162aaafe58bc64a77ccf4f5f963ca0b7`
`Qwen3.8-27B-DFlash2-Q4_K_M.gguf`	Q4_K_M	1.06 GiB	1,143,006,592	4.71	`e44b99d7f4bce8ad5b573190c95e20678613fec9c9f455b073b6d9ef99c241e1`
`Qwen3.8-27B-DFlash2-Q4_K_S.gguf`	Q4_K_S	1.02 GiB	1,096,967,552	4.51	`55bf2541d4d7af2c51153aafaab981d8e1caabeb29fff4923179951b65111bda`

There is no Q6_K_M in llama.cpp; Q6_K is the 6-bit K-quant. If you want the smallest usable draft, take Q4_K_S. If you want the draft to cost as little accuracy as possible, take Q6_K or BF16.

Model info

Property	Value
Parameters	1,924,404,480 (about 1.92B)
Layers	5 (sliding attention, window 2048)
Hidden size	5120
Attention heads	32 query, 8 key/value (head dim 128)
Feed forward	17408
Context length	262,144
Vocabulary	248,320
Speculation block size	8 (7 draft tokens per verification step)
Convolution	kernel 2, group 16
Selector	rank 256, top-k 16
Target layers	5, 19, 33, 47, 61
RoPE theta	10,000,000

Source model: z-lab/Qwen3.8-27B-DFlash2, commit 50307d4c4cde6860d4eee73e2547cd786fe8e8a4, Apache-2.0.

Requirements

DFlash 2 support in llama.cpp is not merged into master yet (as of 19 August 2026). Use a build from PR #27342 or a later master commit once it lands. The files here were produced and validated with a build of that PR merged at commit 5c094f254. Official release binaries will reject the dflash architecture keys until the PR ships.

Usage with llama.cpp

Serve the target model with this draft (DFlash 2 is detected from the checkpoint, no extra flag needed):

./llama-server -m Qwen3.8-27B-Q4_K_M.gguf \
    -md Qwen3.8-27B-DFlash2-Q6_K.gguf \
    --spec-type draft-dflash --spec-draft-n-max 7 -fa on --jinja

The --spec-draft-n-max 7 matches the block size of 8, which gives 7 draft tokens per verification step. Generation without a server:

./llama-cli -m Qwen3.8-27B-Q4_K_M.gguf \
    -md Qwen3.8-27B-DFlash2-Q6_K.gguf \
    --spec-type draft-dflash --spec-draft-n-max 7 -fa on --jinja \
    -p "Hello" -n 128

The draft model needs roughly 1.0 to 3.6 GiB of memory depending on the quant, on top of the target model.

How these files were made

# tokenizer comes from the target model, so --target-model-dir is required
python convert_hf_to_gguf.py z-lab/Qwen3.8-27B-DFlash2 \
    --target-model-dir Qwen/Qwen3.8-27B \
    --outtype bf16 --outfile Qwen3.8-27B-DFlash2-BF16.gguf

./llama-quantize Qwen3.8-27B-DFlash2-BF16.gguf Qwen3.8-27B-DFlash2-Q6_K.gguf Q6_K
./llama-quantize Qwen3.8-27B-DFlash2-BF16.gguf Qwen3.8-27B-DFlash2-Q4_K_M.gguf Q4_K_M
./llama-quantize Qwen3.8-27B-DFlash2-BF16.gguf Qwen3.8-27B-DFlash2-Q4_K_S.gguf Q4_K_S

All 81 tensors load in every file. The quantizer reports these figures:

Quant	Model size	Quantized size
BF16	3671.02 MiB	3671.02 MiB
Q6_K	3671.02 MiB	1506.29 MiB
Q4_K_M	3671.02 MiB	1079.61 MiB
Q4_K_S	3671.02 MiB	1035.71 MiB

What DFlash 2 buys you

Numbers below come from the z-lab model card. They measured SGLang on one NVIDIA H200, FlashAttention 3, block size 8, temperature 1.0, top-p 0.95, top-k 20. Acceptance length is the per-request mean of completion tokens divided by verification steps; higher is better. All speculative methods propose 7 tokens per step.

Task	Qwen3.8 built-in MTP	DFlash 2
GSM8K	5.02	5.46
MATH-500	4.72	5.28
HumanEval	3.91	4.39
MBPP	3.99	4.79
MT-Bench	3.74	4.10

At concurrency 1, throughput speedup over plain autoregressive decoding ranged from 2.67x to 3.43x across those five tasks. The llama.cpp PR reports 5.46 acceptance on GSM8K at Q4_K_M on an Apple M5 Pro with 64 GB, using the first 8 GSM8K problems.

Sources

Draft model: z-lab/Qwen3.8-27B-DFlash2 (mirror of incoai/Qwen3.8-27B-DFlash2)
Target model: Qwen/Qwen3.8-27B
llama.cpp DFlash 2 support: PR #27342
DFlash 2 blog: inco.ai/blog/dflash2, code: github.com/z-lab/dflash

License: Apache-2.0, same as the source model. The tokenizer in these GGUFs comes from the target model, Qwen/Qwen3.8-27B, also Apache-2.0.

Downloads last month: 331

GGUF

Model size

2B params

Architecture

dflash

Hardware compatibility

4-bit

6-bit

16-bit

Model tree for Akicou/Qwen3.8-27B-DFlash2-GGUF

Base model

Qwen/Qwen3.8-27B

Quantized

(1648)

this model