Instructions to use unsloth/Qwen3.8-Flash-Next-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 unsloth/Qwen3.8-Flash-Next-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 unsloth/Qwen3.8-Flash-Next-GGUF:UD-Q4_K_XL
# Run inference directly in the terminal:
llama cli -hf unsloth/Qwen3.8-Flash-Next-GGUF:UD-Q4_K_XL

Install from WinGet (Windows)

winget install llama.cpp
# Start a local OpenAI-compatible server with a web UI:
llama serve -hf unsloth/Qwen3.8-Flash-Next-GGUF:UD-Q4_K_XL
# Run inference directly in the terminal:
llama cli -hf unsloth/Qwen3.8-Flash-Next-GGUF:UD-Q4_K_XL

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 unsloth/Qwen3.8-Flash-Next-GGUF:UD-Q4_K_XL
# Run inference directly in the terminal:
./llama-cli -hf unsloth/Qwen3.8-Flash-Next-GGUF:UD-Q4_K_XL

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 unsloth/Qwen3.8-Flash-Next-GGUF:UD-Q4_K_XL
# Run inference directly in the terminal:
./build/bin/llama-cli -hf unsloth/Qwen3.8-Flash-Next-GGUF:UD-Q4_K_XL

Use Docker

docker model run hf.co/unsloth/Qwen3.8-Flash-Next-GGUF:UD-Q4_K_XL

LM Studio
Jan

vLLM

How to use unsloth/Qwen3.8-Flash-Next-GGUF with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "unsloth/Qwen3.8-Flash-Next-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": "unsloth/Qwen3.8-Flash-Next-GGUF",
		"messages": [
			{
				"role": "user",
				"content": [
					{
						"type": "text",
						"text": "Describe this image in one sentence."
					},
					{
						"type": "image_url",
						"image_url": {
							"url": "https://cdn.britannica.com/61/93061-050-99147DCE/Statue-of-Liberty-Island-New-York-Bay.jpg"
						}
					}
				]
			}
		]
	}'

Use Docker

docker model run hf.co/unsloth/Qwen3.8-Flash-Next-GGUF:UD-Q4_K_XL

Ollama
How to use unsloth/Qwen3.8-Flash-Next-GGUF with Ollama:
```
ollama run hf.co/unsloth/Qwen3.8-Flash-Next-GGUF:UD-Q4_K_XL
```
Unsloth Desktop

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

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf unsloth/Qwen3.8-Flash-Next-GGUF:UD-Q4_K_XL

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": "unsloth/Qwen3.8-Flash-Next-GGUF:UD-Q4_K_XL"
        }
      ]
    }
  }
}

Run Pi

# Start Pi in your project directory:
pi

Docker Model Runner
How to use unsloth/Qwen3.8-Flash-Next-GGUF with Docker Model Runner:
```
docker model run hf.co/unsloth/Qwen3.8-Flash-Next-GGUF:UD-Q4_K_XL
```

Lemonade

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

Pull the model

# Download Lemonade from https://lemonade-server.ai/
lemonade pull unsloth/Qwen3.8-Flash-Next-GGUF:UD-Q4_K_XL

Run and chat with the model

lemonade run user.Qwen3.8-Flash-Next-GGUF-UD-Q4_K_XL

List all available models

lemonade list

Hermes Agent

How to use unsloth/Qwen3.8-Flash-Next-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 unsloth/Qwen3.8-Flash-Next-GGUF:UD-Q4_K_XL

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 unsloth/Qwen3.8-Flash-Next-GGUF:UD-Q4_K_XL

Run Hermes

hermes

Atomic Chat

OpenClaw

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

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf unsloth/Qwen3.8-Flash-Next-GGUF:UD-Q4_K_XL

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 "unsloth/Qwen3.8-Flash-Next-GGUF:UD-Q4_K_XL" \
  --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"

Can the 51B n-gram table be kept on NVMe/SSD?

#44

by NamerPRO - opened 14 days ago

Discussion

NamerPRO

14 days ago

•

edited 14 days ago

Does this Qwen3.8-Flash-Next UD-Q4_K_XL GGUF support the same n-gram SSD offloading approach as AtomicChat's version - i.e. keeping the 51B n-gram embedding table in a separate GGUF file, memory-mapped on NVMe, with only the required rows loaded on demand?

If so, is there already a supported way to run it with 64 GB RAM?

Interpause

14 days ago

https://huggingface.co/unsloth/Qwen3.8-Flash-Next-GGUF/discussions/35#6a9168ae2cc54f730719e80c

I used this guy's weights where he ripped out the ngram table into its own gguf shard, now ive 20 tps on 64GB of DDR4 + RTX 4090 running a 151GB model

NamerPRO

14 days ago

https://huggingface.co/unsloth/Qwen3.8-Flash-Next-GGUF/discussions/35#6a9168ae2cc54f730719e80c

I used this guy's weights where he ripped out the ngram table into its own gguf shard, now ive 20 tps on 64GB of DDR4 + RTX 4090 running a 151GB model

Thank you very much. I will defenitly try it!

DanLar75

14 days ago

I think key to maxing token rate when you offload is to offload in blocks intelligently so that the data most accessed lands on your fastest storage, and the least accessed on your slowest. Just offloading it all to SSD is a terrible idea.

There are solid examples on reddit.

That said. If you are not on Unified memory, you should probably be running 3.8 27B anyway.

Interpause

14 days ago

I think key to maxing token rate when you offload is to offload in blocks intelligently so that the data most accessed lands on your fastest storage, and the least accessed on your slowest. Just offloading it all to SSD is a terrible idea.

There are solid examples on reddit.

That said. If you are not on Unified memory, you should probably be running 3.8 27B anyway.

as long as you leave some RAM leftover, mmap should naturally handle that?

DanLar75

14 days ago

I think key to maxing token rate when you offload is to offload in blocks intelligently so that the data most accessed lands on your fastest storage, and the least accessed on your slowest. Just offloading it all to SSD is a terrible idea.

There are solid examples on reddit.

That said. If you are not on Unified memory, you should probably be running 3.8 27B anyway.

as long as you leave some RAM leftover, mmap should naturally handle that?

I honestly don't know. But the testing I have done on a Threadripper Pro 7965, 128GB DDR5-6000, with 2 x A5000 24GB Nvlink gives me around 70 tokens/sec at Q4, while optimized ram-offload 3.8 Flash-next maybe 15 or so.

krzysztofma

13 days ago

•

edited 13 days ago

I think key to maxing token rate when you offload is to offload in blocks intelligently so that the data most accessed lands on your fastest storage, and the least accessed on your slowest. Just offloading it all to SSD is a terrible idea.

There are solid examples on reddit.

That said. If you are not on Unified memory, you should probably be running 3.8 27B anyway.

as long as you leave some RAM leftover, mmap should naturally handle that?

I honestly don't know. But the testing I have done on a Threadripper Pro 7965, 128GB DDR5-6000, with 2 x A5000 24GB Nvlink gives me around 70 tokens/sec at Q4, while optimized ram-offload 3.8 Flash-next maybe 15 or so.

Isn’t that because of different reasons and not just n-gram becoming bottleneck ? E.g. with your high bandwidth 4-8 channel DDR5 the expert layers are streamed to VRAM , so no wonder it’s fast as long as it fits etc. Perhaps the n-gram ram-offload is not that optimized , or at least not for your setup.

Upload images, audio, and videos by dragging in the text input, pasting, or clicking here.

Tap or paste here to upload images

· Sign up or log in to comment