Instructions to use unsloth/gpt-oss-120b-GGUF with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use unsloth/gpt-oss-120b-GGUF with Transformers:

# Use a pipeline as a high-level helper
from transformers import pipeline

pipe = pipeline("text-generation", model="unsloth/gpt-oss-120b-GGUF")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("unsloth/gpt-oss-120b-GGUF")
model = AutoModelForCausalLM.from_pretrained("unsloth/gpt-oss-120b-GGUF", device_map="auto")

Notebooks
Google Colab
Kaggle
Local Apps Settings

llama.cpp

How to use unsloth/gpt-oss-120b-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/gpt-oss-120b-GGUF:UD-Q4_K_XL
# Run inference directly in the terminal:
llama cli -hf unsloth/gpt-oss-120b-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/gpt-oss-120b-GGUF:UD-Q4_K_XL
# Run inference directly in the terminal:
llama cli -hf unsloth/gpt-oss-120b-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/gpt-oss-120b-GGUF:UD-Q4_K_XL
# Run inference directly in the terminal:
./llama-cli -hf unsloth/gpt-oss-120b-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/gpt-oss-120b-GGUF:UD-Q4_K_XL
# Run inference directly in the terminal:
./build/bin/llama-cli -hf unsloth/gpt-oss-120b-GGUF:UD-Q4_K_XL

Use Docker

docker model run hf.co/unsloth/gpt-oss-120b-GGUF:UD-Q4_K_XL

LM Studio
Jan

vLLM

How to use unsloth/gpt-oss-120b-GGUF with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "unsloth/gpt-oss-120b-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/gpt-oss-120b-GGUF",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/unsloth/gpt-oss-120b-GGUF:UD-Q4_K_XL

SGLang

How to use unsloth/gpt-oss-120b-GGUF with SGLang:

Install from pip and serve model

# Install SGLang from pip:
pip install sglang
# Start the SGLang server:
python3 -m sglang.launch_server \
    --model-path "unsloth/gpt-oss-120b-GGUF" \
    --host 0.0.0.0 \
    --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "unsloth/gpt-oss-120b-GGUF",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker images

docker run --gpus all \
    --shm-size 32g \
    -p 30000:30000 \
    -v ~/.cache/huggingface:/root/.cache/huggingface \
    --env "HF_TOKEN=<secret>" \
    --ipc=host \
    lmsysorg/sglang:latest \
    python3 -m sglang.launch_server \
        --model-path "unsloth/gpt-oss-120b-GGUF" \
        --host 0.0.0.0 \
        --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "unsloth/gpt-oss-120b-GGUF",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Ollama
How to use unsloth/gpt-oss-120b-GGUF with Ollama:
```
ollama run hf.co/unsloth/gpt-oss-120b-GGUF:UD-Q4_K_XL
```

Unsloth Studio

How to use unsloth/gpt-oss-120b-GGUF with Unsloth Studio:

Install Unsloth Studio (macOS, Linux, WSL)

curl -fsSL https://unsloth.ai/install.sh | sh
# Run unsloth studio
unsloth studio -H 0.0.0.0 -p 8888
# Then open http://localhost:8888 in your browser
# Search for unsloth/gpt-oss-120b-GGUF to start chatting

Install Unsloth Studio (Windows)

irm https://unsloth.ai/install.ps1 | iex
# Run unsloth studio
unsloth studio -H 0.0.0.0 -p 8888
# Then open http://localhost:8888 in your browser
# Search for unsloth/gpt-oss-120b-GGUF to start chatting

Using HuggingFace Spaces for Unsloth

# No setup required
# Open https://huggingface.co/spaces/unsloth/studio in your browser
# Search for unsloth/gpt-oss-120b-GGUF to start chatting

How to use unsloth/gpt-oss-120b-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/gpt-oss-120b-GGUF:UD-Q4_K_XL

Configure the model in Pi

# Install Pi:
npm install -g @mariozechner/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/gpt-oss-120b-GGUF:UD-Q4_K_XL"
        }
      ]
    }
  }
}

Run Pi

# Start Pi in your project directory:
pi

Docker Model Runner
How to use unsloth/gpt-oss-120b-GGUF with Docker Model Runner:
```
docker model run hf.co/unsloth/gpt-oss-120b-GGUF:UD-Q4_K_XL
```

Lemonade

How to use unsloth/gpt-oss-120b-GGUF with Lemonade:

Pull the model

# Download Lemonade from https://lemonade-server.ai/
lemonade pull unsloth/gpt-oss-120b-GGUF:UD-Q4_K_XL

Run and chat with the model

lemonade run user.gpt-oss-120b-GGUF-UD-Q4_K_XL

List all available models

lemonade list

Hermes Agent

How to use unsloth/gpt-oss-120b-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/gpt-oss-120b-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/gpt-oss-120b-GGUF:UD-Q4_K_XL

Run Hermes

hermes

Atomic Chat

OpenClaw

How to use unsloth/gpt-oss-120b-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/gpt-oss-120b-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/gpt-oss-120b-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"

GGUF is very slow for some reason

#12

by ineersa - opened Aug 24, 2025

Discussion

ineersa

Aug 24, 2025

Hi there, thanks for great job!

I was trying to utilize 2 GPUs and swapped to this GGUFs from LM Studio MXFP4, and it's surprisingly slow.

I guess something is wrong with my setup but i can't figure it out.

I'm running llama.cpp:

llama-server --device CUDA0 --model ~/models/unsloth/gpt-oss-120b/Q8_0/gpt-oss-120b-Q8_0-00001-of-00002.gguf \
  --host 0.0.0.0 \
  --port 8052 \
  --jinja \
  --ctx-size 65536 \
  --threads 1 \
  --n-cpu-moe 30 \
  --flash-attn \
  --temp 1.0 \
  --top-p 1.0 \
  --min-p 0.0 \
  --top-k 0 \
  --n-gpu-layers 99 \
  --chat-template-kwargs '{"builtin_tools":["python", "browser"], "reasoning_effort":"high"}'

Test with single GPU first.
Q8_0 GGUF gives me 6 tk/s, when MXFP4 from LM Studio gives around 25 tk/s with exaclty same run command.

I was trying with both GPUs with -ot ".ffn_(up|down)_exps.=CPU" but it's also around 7 tk/s.

llama-server version: 6257 (b1afcab8)

cmake -B build -DGGML_CUDA=ON -DGGML_CUDA_FA_ALL_QUANTS=ON -DGGML_CUDA_F16=ON -DCURL_LIBRARY=/usr/lib/x86_64-linux-gnu/libcurl.so -DGGML_CUDA_MMQ=ON

Could you help me to understand what is not working properly?

ineersa

Aug 25, 2025

https://docs.unsloth.ai/basics/gpt-oss-how-to-run-and-fine-tune#run-gpt-oss-120b

Inside this guide there is an option --threads 1 which pretty much same as default.
But if i don't pass --threads 1 I'm getting 22-23 tokens/sec, when i pass --threads 10 I'm getting 23-24 tokens/sec, but if i pass --threads 1 I get 6 tokens/sec.

That's some black magic going on here.

Now it works, Q8_0 runs at pretty much same speed as MXFP4.

llama-server --device CUDA0,CUDA1 \
  --model ~/models/unsloth/gpt-oss-120b/Q8_0/gpt-oss-120b-Q8_0-00001-of-00002.gguf \
  --host 0.0.0.0 \
  --port 8052 \
  --jinja \
  --threads 10 \
  --ctx-size 65536 \
  --batch-size 4096 \
  --ubatch-size 2048 \
  -ot ".ffn_(up|down)_exps.=CPU" \
  --flash-attn \
  --temp 1.0 \
  --top-p 1.0 \
  --top-k 0 \
  --n-gpu-layers 999 \
  --chat-template-kwargs '{"builtin_tools":["python", "browser"], "reasoning_effort":"high"}'

Gives around 26 tokens/sec, difference is minimal with 1 GPU, so I guess i will stick to 1 GPU and run other models on 2nd one.

Gerald001

Feb 28

•

edited Feb 28

hi @ineersa will the following command also work for https://huggingface.co/unsloth/gpt-oss-safeguard-20b-GGUF ?

llama-server --device CUDA0,CUDA1
--model ~/models/unsloth/gpt-oss-safeguard-20b-GGUF /gpt-oss-safeguard-20b-Q2_K_L.gguf
--host 0.0.0.0
--port 8052
--jinja
--threads 10
--ctx-size 65536
--batch-size 4096
--ubatch-size 2048
-ot ".ffn_(up|down)_exps.=CPU"
--flash-attn
--temp 1.0
--top-p 1.0
--top-k 0
--n-gpu-layers 999
--chat-template-kwargs '{"builtin_tools":["python", "browser"], "reasoning_effort":"high"}'

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