Instructions to use shafire/Zero-Qwen3-8B-OpenZero-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 shafire/Zero-Qwen3-8B-OpenZero-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 shafire/Zero-Qwen3-8B-OpenZero-GGUF:F16
# Run inference directly in the terminal:
llama cli -hf shafire/Zero-Qwen3-8B-OpenZero-GGUF:F16

Install from WinGet (Windows)

winget install llama.cpp
# Start a local OpenAI-compatible server with a web UI:
llama serve -hf shafire/Zero-Qwen3-8B-OpenZero-GGUF:F16
# Run inference directly in the terminal:
llama cli -hf shafire/Zero-Qwen3-8B-OpenZero-GGUF:F16

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 shafire/Zero-Qwen3-8B-OpenZero-GGUF:F16
# Run inference directly in the terminal:
./llama-cli -hf shafire/Zero-Qwen3-8B-OpenZero-GGUF:F16

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 shafire/Zero-Qwen3-8B-OpenZero-GGUF:F16
# Run inference directly in the terminal:
./build/bin/llama-cli -hf shafire/Zero-Qwen3-8B-OpenZero-GGUF:F16

Use Docker

docker model run hf.co/shafire/Zero-Qwen3-8B-OpenZero-GGUF:F16

LM Studio
Jan

vLLM

How to use shafire/Zero-Qwen3-8B-OpenZero-GGUF with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "shafire/Zero-Qwen3-8B-OpenZero-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": "shafire/Zero-Qwen3-8B-OpenZero-GGUF",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/shafire/Zero-Qwen3-8B-OpenZero-GGUF:F16

Ollama
How to use shafire/Zero-Qwen3-8B-OpenZero-GGUF with Ollama:
```
ollama run hf.co/shafire/Zero-Qwen3-8B-OpenZero-GGUF:F16
```

Unsloth Studio

How to use shafire/Zero-Qwen3-8B-OpenZero-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 shafire/Zero-Qwen3-8B-OpenZero-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 shafire/Zero-Qwen3-8B-OpenZero-GGUF to start chatting

Using HuggingFace Spaces for Unsloth

# No setup required
# Open https://huggingface.co/spaces/unsloth/studio in your browser
# Search for shafire/Zero-Qwen3-8B-OpenZero-GGUF to start chatting

How to use shafire/Zero-Qwen3-8B-OpenZero-GGUF with Pi:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf shafire/Zero-Qwen3-8B-OpenZero-GGUF:F16

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": "shafire/Zero-Qwen3-8B-OpenZero-GGUF:F16"
        }
      ]
    }
  }
}

Run Pi

# Start Pi in your project directory:
pi

Hermes Agent new

How to use shafire/Zero-Qwen3-8B-OpenZero-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 shafire/Zero-Qwen3-8B-OpenZero-GGUF:F16

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 shafire/Zero-Qwen3-8B-OpenZero-GGUF:F16

Run Hermes

hermes

Atomic Chat new

OpenClaw new

How to use shafire/Zero-Qwen3-8B-OpenZero-GGUF with OpenClaw:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf shafire/Zero-Qwen3-8B-OpenZero-GGUF:F16

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 "shafire/Zero-Qwen3-8B-OpenZero-GGUF:F16" \
  --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"

Docker Model Runner
How to use shafire/Zero-Qwen3-8B-OpenZero-GGUF with Docker Model Runner:
```
docker model run hf.co/shafire/Zero-Qwen3-8B-OpenZero-GGUF:F16
```

Lemonade

How to use shafire/Zero-Qwen3-8B-OpenZero-GGUF with Lemonade:

Pull the model

# Download Lemonade from https://lemonade-server.ai/
lemonade pull shafire/Zero-Qwen3-8B-OpenZero-GGUF:F16

Run and chat with the model

lemonade run user.Zero-Qwen3-8B-OpenZero-GGUF-F16

List all available models

lemonade list

ZERO Qwen3-8B OpenZero — Standalone Agentic GGUF

ONE FILE. NO ADAPTER. NO BASE-MODEL HUNT.
Local coding, research and agentic AI built for OpenZero.

Zero Qwen3-8B OpenZero is a fully merged, standalone GGUF model for local research, coding, debugging, tool use and autonomous agent workflows. Download one GGUF and run it directly. The OpenZero fine-tune is already fused into the model.

Zero does not wait for permission to think. It verifies, acts and reports.

OpenZero 7.1 demo

▶ Watch “OpenZero 7.1 — 16 Agents, 1 LLM” on YouTube

The demo shows the local-first OpenZero control panel running a 16-agent autonomy configuration around one local LLM.

OpenZero local model family

This release is one of the three local model choices shown in the current OpenZero and ZeroThink interfaces:

Model	Local runtime alias	Hugging Face
OpenZero Gemma 4 E4B	`openzerogemma:latest`	Zero-Gemma4-E4B-OpenZero-GGUF
OpenZero Qwen3 8B Q5_K_M	`zero-qwen3-q5:latest`	this repository
OpenZero Qwen3 8B F16	`zero-qwen3-f16:latest`	this repository

Run directly from Hugging Face with Ollama:

ollama run hf.co/shafire/Zero-Qwen3-8B-OpenZero-GGUF:Q5_K_M
ollama run hf.co/shafire/Zero-Gemma4-E4B-OpenZero-GGUF

OpenZero 7.1 can select the installed aliases locally. ZeroThink can route through its OpenZero Local provider when connected to an OpenZero node. Alias availability depends on which quantizations are installed on the machine.

OpenZero in action

_{OpenZero 7.1 — Ultra autonomy with the 16-agent setting.}	_{ZeroThink Studio — the three current OpenZero local model choices.}
_{Local Agent Zero response inside the OpenZero control panel.}
_{OpenZero agentic workflow and local privacy controls.}

Download this model

File	Size	Best for
`Zero-Qwen3-8B-OpenZero-Q5_K_M.gguf`	5.45 GiB	Recommended local release: strong quality and practical CPU/RAM use
`Zero-Qwen3-8B-OpenZero-FUSED-F16.gguf`	13.82 GiB	High-precision fused build for advanced users and further quantization

Recommended: Zero-Qwen3-8B-OpenZero-Q5_K_M.gguf

Standalone model: yes
Separate adapter required: no
Separate base model required: no
llama.cpp compatible: yes
OpenZero compatible: yes
CPU generation tested: yes

What Zero is built for

Agentic coding: implementation, debugging, code review and test planning
Deep research: evidence-led synthesis, uncertainty tracking and verification
Autonomous workflows: planning, tool selection, execution and result checks
Local AI: private inference through llama.cpp, Ollama and compatible runtimes
OpenZero: OpenAI-compatible local serving for multi-agent and research systems
Zero persona: direct, dry, skeptical and resistant to unsupported claims

Run with llama.cpp

llama-cli \
  -m Zero-Qwen3-8B-OpenZero-Q5_K_M.gguf \
  --jinja -c 8192 -t 8 \
  --temp 0.6 --top-k 20 --top-p 0.95

Start an OpenAI-compatible endpoint for OpenZero:

llama-server \
  -m Zero-Qwen3-8B-OpenZero-Q5_K_M.gguf \
  --jinja -c 8192 -t 8 \
  --host 127.0.0.1 --port 8080

Use http://127.0.0.1:8080/v1 as the local OpenAI-compatible API base.

Qwen3 supports /no_think for fast tool loops and /think for difficult research or coding.

Run with Ollama

Create Modelfile beside the GGUF:

FROM ./Zero-Qwen3-8B-OpenZero-Q5_K_M.gguf
PARAMETER temperature 0.6
PARAMETER top_p 0.95
PARAMETER num_ctx 8192

ollama create zero-qwen3 -f Modelfile
ollama run zero-qwen3

Real CPU benchmark

Measured locally with llama-bench from llama.cpp b10107:

Hardware / test	Prompt processing	Token generation
Intel Core i7-2600, 4C/8T, CPU-only, 8 threads, pp64 / tg16	4.41 tok/s	2.45 tok/s

This is deliberately old CPU hardware. Results are a reproducible deployment reference, not a claim of universal speed. Modern CPUs and GPU offload should perform differently.

Verified release

Base architecture: Qwen/Qwen3-8B
Full fine-tuning run: 2,033 curated OpenZero examples
Train / validation split: 2,025 / 8
Final validation loss: 1.293387
Final validation mean token accuracy: 0.846873
Fusion: 252 LoRA-targeted tensors merged; 399 tensors written
Q5_K_M SHA-256: 390464f750b5cb53da298848adc05839c1fd40404a74cd5f800cad9612d17d59
CPU load and text-generation smoke test: PASS

The training material is already represented in the merged weights. Users do not need the training dataset, a PEFT archive or a LoRA adapter to run Zero.

Practical notes

Start with an 8K context on a 16 GiB system and raise it only after measuring available memory.
This is a text-generation model. Tool execution belongs in the surrounding agent runtime.
The 512-token fine-tuning window strengthened targeted behavior; it does not redefine the base model's entire long-context performance.
Validate outputs before using them for security, medical, legal or financial decisions.