Instructions to use oOundefinedOo/Qwen3-0.6B-Visual-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 oOundefinedOo/Qwen3-0.6B-Visual-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 oOundefinedOo/Qwen3-0.6B-Visual-GGUF:Q4_K_M
# Run inference directly in the terminal:
llama cli -hf oOundefinedOo/Qwen3-0.6B-Visual-GGUF:Q4_K_M

Install from WinGet (Windows)

winget install llama.cpp
# Start a local OpenAI-compatible server with a web UI:
llama serve -hf oOundefinedOo/Qwen3-0.6B-Visual-GGUF:Q4_K_M
# Run inference directly in the terminal:
llama cli -hf oOundefinedOo/Qwen3-0.6B-Visual-GGUF:Q4_K_M

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 oOundefinedOo/Qwen3-0.6B-Visual-GGUF:Q4_K_M
# Run inference directly in the terminal:
./llama-cli -hf oOundefinedOo/Qwen3-0.6B-Visual-GGUF:Q4_K_M

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 oOundefinedOo/Qwen3-0.6B-Visual-GGUF:Q4_K_M
# Run inference directly in the terminal:
./build/bin/llama-cli -hf oOundefinedOo/Qwen3-0.6B-Visual-GGUF:Q4_K_M

Use Docker

docker model run hf.co/oOundefinedOo/Qwen3-0.6B-Visual-GGUF:Q4_K_M

LM Studio
Jan

vLLM

How to use oOundefinedOo/Qwen3-0.6B-Visual-GGUF with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "oOundefinedOo/Qwen3-0.6B-Visual-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": "oOundefinedOo/Qwen3-0.6B-Visual-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/oOundefinedOo/Qwen3-0.6B-Visual-GGUF:Q4_K_M

Ollama
How to use oOundefinedOo/Qwen3-0.6B-Visual-GGUF with Ollama:
```
ollama run hf.co/oOundefinedOo/Qwen3-0.6B-Visual-GGUF:Q4_K_M
```
Unsloth Desktop

How to use oOundefinedOo/Qwen3-0.6B-Visual-GGUF with Pi:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf oOundefinedOo/Qwen3-0.6B-Visual-GGUF:Q4_K_M

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": "oOundefinedOo/Qwen3-0.6B-Visual-GGUF:Q4_K_M"
        }
      ]
    }
  }
}

Run Pi

# Start Pi in your project directory:
pi

Docker Model Runner
How to use oOundefinedOo/Qwen3-0.6B-Visual-GGUF with Docker Model Runner:
```
docker model run hf.co/oOundefinedOo/Qwen3-0.6B-Visual-GGUF:Q4_K_M
```

Lemonade

How to use oOundefinedOo/Qwen3-0.6B-Visual-GGUF with Lemonade:

Pull the model

# Download Lemonade from https://lemonade-server.ai/
lemonade pull oOundefinedOo/Qwen3-0.6B-Visual-GGUF:Q4_K_M

Run and chat with the model

lemonade run user.Qwen3-0.6B-Visual-GGUF-Q4_K_M

List all available models

lemonade list

Hermes Agent

How to use oOundefinedOo/Qwen3-0.6B-Visual-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 oOundefinedOo/Qwen3-0.6B-Visual-GGUF:Q4_K_M

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 oOundefinedOo/Qwen3-0.6B-Visual-GGUF:Q4_K_M

Run Hermes

hermes

Atomic Chat

OpenClaw

How to use oOundefinedOo/Qwen3-0.6B-Visual-GGUF with OpenClaw:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf oOundefinedOo/Qwen3-0.6B-Visual-GGUF:Q4_K_M

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 "oOundefinedOo/Qwen3-0.6B-Visual-GGUF:Q4_K_M" \
  --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"

Qwen3-0.6B-Visual-GGUF

GGUF files of Qwen3-0.6B-VL — Qwen3-0.6B that can see images, with thinking mode and tool calling — for llama.cpp, Ollama, LM Studio and other GGUF apps. Runs comfortably on CPU.

File	Size
`Qwen3-0.6B-Visual-Q4_K_M.gguf`	378 MB	recommended, smallest
`Qwen3-0.6B-Visual-Q8_0.gguf`	610 MB	near-lossless
`Qwen3-0.6B-Visual-BF16.gguf`	1.12 GB	unquantized
`mmproj-Qwen3-0.6B-Visual-F16.gguf`	167 MB	vision encoder, needed for images

Ollama

ollama run hf.co/oOundefinedOo/Qwen3-0.6B-Visual-GGUF:Q4_K_M

Vision, thinking and tools work out of the box (the repo ships a chat template). Drag an image into the chat or pass it in the API images field. Thinking is on by default as in Qwen3; turn it off with --think=false or /set nothink. To install from local files instead, download the files and run ollama create qwen3-0.6b-visual -f Modelfile.

LM Studio

Search for Qwen3-0.6B-Visual-GGUF and download a quantization — the mmproj file is picked up automatically.

llama.cpp

llama-server -hf oOundefinedOo/Qwen3-0.6B-Visual-GGUF:Q4_K_M --jinja

Or with local files: llama-mtmd-cli -m Qwen3-0.6B-Visual-Q4_K_M.gguf --mmproj mmproj-Qwen3-0.6B-Visual-F16.gguf --image photo.jpg -p "Describe this image.". With llama-server, pass "chat_template_kwargs": {"enable_thinking": false} to switch thinking off.

Quality

Measured on 300 POPE questions (object hallucination, accuracy) against the original transformers model on the same questions:

	transformers	BF16	Q8_0	Q4_K_M
Accuracy	77.0	78.0	78.0	75.0

Tips: for tool calling, thinking mode is noticeably more reliable. On simple visual questions ("is there a dog?") the model usually answers directly even with thinking on — that is more accurate for a model this size.

The vision encoder runs at 224×224, so small text and fine details are not visible. See the main model card for details, and try the demo Space.

Downloads last month: 246

GGUF

Model size

0.6B params

Architecture

qwen3

Hardware compatibility

4-bit

8-bit

16-bit

Model tree for oOundefinedOo/Qwen3-0.6B-Visual-GGUF

Base model

Qwen/Qwen3-0.6B-Base

Finetuned

Qwen/Qwen3-0.6B

Finetuned

oOundefinedOo/Qwen3-0.6B-VL

Quantized

(1)

this model