Instructions to use MoziLabs/Mozi-2b with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use MoziLabs/Mozi-2b with Transformers:

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

pipe = pipeline("text-generation", model="MoziLabs/Mozi-2b")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# pip install -U transformers accelerate
# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("MoziLabs/Mozi-2b")
model = AutoModelForCausalLM.from_pretrained("MoziLabs/Mozi-2b", device_map="auto")
messages = [
    {"role": "user", "content": "Who are you?"},
]
inputs = tokenizer.apply_chat_template(
	messages,
	add_generation_prompt=True,
	tokenize=True,
	return_dict=True,
	return_tensors="pt",
).to(model.device)

outputs = model.generate(**inputs, max_new_tokens=256)
print(tokenizer.decode(outputs[0][inputs["input_ids"].shape[-1]:]))

Notebooks
Google Colab
Kaggle
Local Apps Settings

llama.cpp

How to use MoziLabs/Mozi-2b 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 MoziLabs/Mozi-2b:Q8_0
# Run inference directly in the terminal:
llama cli -hf MoziLabs/Mozi-2b:Q8_0

Install from WinGet (Windows)

winget install llama.cpp
# Start a local OpenAI-compatible server with a web UI:
llama serve -hf MoziLabs/Mozi-2b:Q8_0
# Run inference directly in the terminal:
llama cli -hf MoziLabs/Mozi-2b:Q8_0

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 MoziLabs/Mozi-2b:Q8_0
# Run inference directly in the terminal:
./build/bin/llama cli -hf MoziLabs/Mozi-2b:Q8_0

Use Docker

docker model run hf.co/MoziLabs/Mozi-2b:Q8_0

Llama new
LM Studio
Jan

vLLM

How to use MoziLabs/Mozi-2b with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "MoziLabs/Mozi-2b"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "MoziLabs/Mozi-2b",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/MoziLabs/Mozi-2b:Q8_0

SGLang

How to use MoziLabs/Mozi-2b 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 "MoziLabs/Mozi-2b" \
    --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": "MoziLabs/Mozi-2b",
		"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 "MoziLabs/Mozi-2b" \
        --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": "MoziLabs/Mozi-2b",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Ollama
How to use MoziLabs/Mozi-2b with Ollama:
```
ollama run hf.co/MoziLabs/Mozi-2b:Q8_0
```
Unsloth Desktop

How to use MoziLabs/Mozi-2b with Pi:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf MoziLabs/Mozi-2b: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": "MoziLabs/Mozi-2b:Q8_0"
        }
      ]
    }
  }
}

Run Pi

# Start Pi in your project directory:
pi

Docker Model Runner
How to use MoziLabs/Mozi-2b with Docker Model Runner:
```
docker model run hf.co/MoziLabs/Mozi-2b:Q8_0
```

Lemonade

How to use MoziLabs/Mozi-2b with Lemonade:

Pull the model

# Download Lemonade from https://lemonade-server.ai/
lemonade pull MoziLabs/Mozi-2b:Q8_0

Run and chat with the model

lemonade run user.Mozi-2b-Q8_0

List all available models

lemonade list

Hermes Agent

How to use MoziLabs/Mozi-2b with Hermes Agent:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf MoziLabs/Mozi-2b: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 MoziLabs/Mozi-2b:Q8_0

Run Hermes

hermes

Atomic Chat

OpenClaw

How to use MoziLabs/Mozi-2b with OpenClaw:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf MoziLabs/Mozi-2b: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 "MoziLabs/Mozi-2b: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"

Mozi-2b

Mozi-2b is a fine-tune of Qwen3.5-2B by MoziLabs. It keeps the base model's thinking mode, but it's trained to reason in a shorter, more adaptive way: quick answers for easy questions and fuller reasoning only when a problem needs it.

How it was made

Teacher: DeepSeek-V4.1-Flash (deepseek-flash) wrote a synthetic dataset of prompts, each with reasoning and a final answer.
Method: LoRA distillation (rank 16, 4-bit QLoRA), 2 epochs, learning rate 2e-4, max sequence length 2048. The reasoning was trained in Qwen3.5's native <think> format, and the LoRA was merged back into the full model.

Files

File	Use
`model.safetensors` + tokenizer/config	Full bf16 weights for 🤗 Transformers
`Mozi-2b-Q8_0.gguf`	Q8_0 GGUF for Ollama / llama.cpp / LM Studio

Usage

Ollama

ollama run hf.co/MoziLabs/Mozi-2b:Q8_0

Transformers

from transformers import AutoModelForCausalLM, AutoTokenizer

tok = AutoTokenizer.from_pretrained("MoziLabs/Mozi-2b")
model = AutoModelForCausalLM.from_pretrained("MoziLabs/Mozi-2b", dtype="auto", device_map="auto")

messages = [{"role": "user", "content": "What is 17 * 23?"}]
text = tok.apply_chat_template(messages, add_generation_prompt=True, enable_thinking=True, tokenize=False)
inputs = tok(text, return_tensors="pt", add_special_tokens=False).to(model.device)
out = model.generate(**inputs, max_new_tokens=2048, temperature=0.6, top_p=0.95, top_k=20, do_sample=True)
print(tok.decode(out[0][inputs["input_ids"].shape[1]:], skip_special_tokens=True))

Limitations

On very hard problems it can still think for a long time before answering. Give it a generous token budget (8k+).
It's a 2B model: expect mistakes on difficult knowledge and math questions, and check anything important.
Text only: the vision part of Qwen3.5-2B is not included, so it can't read images.