Instructions to use drluoto/Qwen3.8-Flash-Next-MTP-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 drluoto/Qwen3.8-Flash-Next-MTP-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 drluoto/Qwen3.8-Flash-Next-MTP-GGUF:Q8_0
# Run inference directly in the terminal:
llama cli -hf drluoto/Qwen3.8-Flash-Next-MTP-GGUF:Q8_0

Install from WinGet (Windows)

winget install llama.cpp
# Start a local OpenAI-compatible server with a web UI:
llama serve -hf drluoto/Qwen3.8-Flash-Next-MTP-GGUF:Q8_0
# Run inference directly in the terminal:
llama cli -hf drluoto/Qwen3.8-Flash-Next-MTP-GGUF:Q8_0

Use pre-built binary

# Download pre-built binary from:
# https://github.com/ggerganov/llama.cpp/releases
# Start a local OpenAI-compatible server with a web UI:
./llama-server -hf drluoto/Qwen3.8-Flash-Next-MTP-GGUF:Q8_0
# Run inference directly in the terminal:
./llama-cli -hf drluoto/Qwen3.8-Flash-Next-MTP-GGUF:Q8_0

Build from source code

git clone https://github.com/ggerganov/llama.cpp.git
cd llama.cpp
cmake -B build
cmake --build build -j --target llama-server llama-cli
# Start a local OpenAI-compatible server with a web UI:
./build/bin/llama-server -hf drluoto/Qwen3.8-Flash-Next-MTP-GGUF:Q8_0
# Run inference directly in the terminal:
./build/bin/llama-cli -hf drluoto/Qwen3.8-Flash-Next-MTP-GGUF:Q8_0

Use Docker

docker model run hf.co/drluoto/Qwen3.8-Flash-Next-MTP-GGUF:Q8_0

LM Studio
Jan
Ollama
How to use drluoto/Qwen3.8-Flash-Next-MTP-GGUF with Ollama:
```
ollama run hf.co/drluoto/Qwen3.8-Flash-Next-MTP-GGUF:Q8_0
```
Unsloth Desktop

How to use drluoto/Qwen3.8-Flash-Next-MTP-GGUF with Pi:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf drluoto/Qwen3.8-Flash-Next-MTP-GGUF:Q8_0

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": "drluoto/Qwen3.8-Flash-Next-MTP-GGUF:Q8_0"
        }
      ]
    }
  }
}

Run Pi

# Start Pi in your project directory:
pi

Docker Model Runner
How to use drluoto/Qwen3.8-Flash-Next-MTP-GGUF with Docker Model Runner:
```
docker model run hf.co/drluoto/Qwen3.8-Flash-Next-MTP-GGUF:Q8_0
```

Lemonade

How to use drluoto/Qwen3.8-Flash-Next-MTP-GGUF with Lemonade:

Pull the model

# Download Lemonade from https://lemonade-server.ai/
lemonade pull drluoto/Qwen3.8-Flash-Next-MTP-GGUF:Q8_0

Run and chat with the model

lemonade run user.Qwen3.8-Flash-Next-MTP-GGUF-Q8_0

List all available models

lemonade list

Hermes Agent

How to use drluoto/Qwen3.8-Flash-Next-MTP-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 drluoto/Qwen3.8-Flash-Next-MTP-GGUF:Q8_0

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 drluoto/Qwen3.8-Flash-Next-MTP-GGUF:Q8_0

Run Hermes

hermes

Atomic Chat

OpenClaw

How to use drluoto/Qwen3.8-Flash-Next-MTP-GGUF with OpenClaw:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf drluoto/Qwen3.8-Flash-Next-MTP-GGUF:Q8_0

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 "drluoto/Qwen3.8-Flash-Next-MTP-GGUF:Q8_0" \
  --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-Flash-Next MTP draft head (Q8_0, ROCm-verified)

Standalone NextN/MTP draft sidecar for Qwen/Qwen3.8-Flash-Next, for use with llama.cpp --spec-type draft-mtp (-md flag).

Full guide, measurements and pitfalls: https://github.com/drluoto/flash-next-strix-halo · Discussion: https://github.com/ggml-org/llama.cpp/discussions/27950

Provenance

Converted directly from the official checkpoint with the --mtp export from llama.cpp PR #27836 plus the whitelist fix (rmonsurate/llama.cpp#1) that keeps the model-level output_hc_norm/_down/_up mixer the loader requires. 37 tensors: the 4B MTP head, shared embeddings/lm_head/norm, and the head mixer.

Only the MTP-relevant tensors (5.2 GB of the 360 GB checkpoint) were fetched, via HTTP range reads on the safetensors shards.

sha256(mtp-egen-Q8_0.gguf) = b9880220df29fc224bbce408c867cd5d9c021263b754033ea624b669e374f4ec

Requirements

Loading a detached head needs crusaderky's loader fix on top of PR #27836. A pre-assembled branch (that + hipCUB TOP_K from #26592): drluoto/llama.cpp branch strix-halo-flash-next.

Measured (Ryzen AI Max+ 395 / Radeon 8060S, ROCm 7.1, target UD-IQ4_XS, greedy)

workload	no spec	`draft-mtp,ngram-mod` + GPU TOP_K
file rewrite @8k ctx	16.8 t/s	47.1 t/s
new code @8k	16.8	31.7
file rewrite @24k	~15	28.6
new code @24k	~15	25.4

Greedy output verified clean at long prompts (the failure mode of earlier community MTP ports was prompt-length-dependent corruption above ~1k tokens).

llama-server -m Qwen3.8-Flash-Next-UD-IQ4_XS-00001-of-00003.gguf \
  -md mtp-Qwen3.8-Flash-Next-Q8_0.gguf \
  --spec-type draft-mtp,ngram-mod --spec-draft-n-max 3 \
  --spec-ngram-mod-n-max 64 --spec-ngram-mod-n-match 24 \
  -lm dio -fa 1 -c 262144 --jinja
# on ROCm < 7.13 add: GGML_CUDA_DISABLE_GRAPHS=1 (rocPRIM capture guard, see #26592)

Downloads last month: -

GGUF

Model size

4B params

Architecture

qwen4exp

Hardware compatibility

8-bit

Inference Providers NEW

This model isn't deployed by any Inference Provider. 🙋 Ask for provider support

Model tree for drluoto/Qwen3.8-Flash-Next-MTP-GGUF

Base model

Qwen/Qwen3.8-Flash-Next

Quantized

(134)

this model