Instructions to use zcyyyds/Qwen3.5-4B-GUI-Edge with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use zcyyyds/Qwen3.5-4B-GUI-Edge with Transformers:

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

pipe = pipeline("image-text-to-text", model="zcyyyds/Qwen3.5-4B-GUI-Edge")
messages = [
    {
        "role": "user",
        "content": [
            {"type": "image", "url": "https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/p-blog/candy.JPG"},
            {"type": "text", "text": "What animal is on the candy?"}
        ]
    },
]
pipe(text=messages)

# pip install -U transformers accelerate
# Load model directly
from transformers import AutoProcessor, AutoModelForMultimodalLM

processor = AutoProcessor.from_pretrained("zcyyyds/Qwen3.5-4B-GUI-Edge")
model = AutoModelForMultimodalLM.from_pretrained("zcyyyds/Qwen3.5-4B-GUI-Edge", device_map="auto")
messages = [
    {
        "role": "user",
        "content": [
            {"type": "image", "url": "https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/p-blog/candy.JPG"},
            {"type": "text", "text": "What animal is on the candy?"}
        ]
    },
]
inputs = processor.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(processor.decode(outputs[0][inputs["input_ids"].shape[-1]:]))

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use zcyyyds/Qwen3.5-4B-GUI-Edge with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "zcyyyds/Qwen3.5-4B-GUI-Edge"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "zcyyyds/Qwen3.5-4B-GUI-Edge",
		"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/zcyyyds/Qwen3.5-4B-GUI-Edge

SGLang

How to use zcyyyds/Qwen3.5-4B-GUI-Edge 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 "zcyyyds/Qwen3.5-4B-GUI-Edge" \
    --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": "zcyyyds/Qwen3.5-4B-GUI-Edge",
		"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 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 "zcyyyds/Qwen3.5-4B-GUI-Edge" \
        --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": "zcyyyds/Qwen3.5-4B-GUI-Edge",
		"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"
						}
					}
				]
			}
		]
	}'

Docker Model Runner
How to use zcyyyds/Qwen3.5-4B-GUI-Edge with Docker Model Runner:
```
docker model run hf.co/zcyyyds/Qwen3.5-4B-GUI-Edge
```

Qwen3.5-4B-GUI-Edge

A 4B GUI agent for phone and Chinese app operation, post-trained from Qwen3.5-4B. It outperforms 8B GUI specialists on AndroidControl and CAGUI.

Results

Step accuracy (%). All models are scored on the same test samples by the same evaluator; baselines use their official prompts.

Model	AndroidControl	CAGUI	CAGUI OCR
GUI-Edge (4B)	81.5	79.5	76.0
UI-Venus-1.5-8B	76.5	72.0	56.0
MAI-UI-8B	69.5	65.5	54.0
GUI-Owl-1.5-4B	71.5	63.5	58.0
UI-TARS-1.5-7B	68.5	64.0	44.0

Method

SFT on 64K samples from AndroidControl, GUI-Odyssey and CAGUI, then GRPO with verifiable action rewards.
Coordinate tokens are 15% of the output but carry 84% of the model's dependence on the image, so the final GRPO stage weights them 4× on high-resolution GroundCUA data.

Quick start

hf download 1ss/Qwen3.5-4B-GUI-Edge --local-dir GUI-Edge
cd GUI-Edge && pip install -r eval/requirements.txt
python infer.py --image screenshot.png --instruction "Open Settings"   # add --backend vllm for vLLM

Output:

<tool_call>
<function=mobile_use>
<parameter=action>click</parameter>
<parameter=coordinate>[500, 300]</parameter>
</function>
</tool_call>

Coordinates are on a 0–999 grid over the screenshot. The LoRA adapter is in adapter/.

Evaluation

python eval/run.py --model . --android-root /data/androidcontrol --cagui-root /data/cagui --output scores

Dataset versions are listed in eval/datasets.json.

License

CC BY-NC 4.0, following the CAGUI data license. The base model Qwen3.5-4B is Apache-2.0 (BASE-MODEL-LICENSE); data attribution is in NOTICE.md.

中文：面向手机与中文 App 操作的 4B GUI 智能体模型，基于 Qwen3.5-4B 后训练，在 AndroidControl 与 CAGUI 上超过 8B GUI 专用模型。

Downloads last month: 28

Safetensors

Model size

5B params

Tensor type

BF16

F32

Model tree for zcyyyds/Qwen3.5-4B-GUI-Edge

Base model

Qwen/Qwen3.5-4B-Base

Finetuned

Qwen/Qwen3.5-4B

Finetuned

(1,259)

this model

zcyyyds
/

Qwen3.5-4B-GUI-Edge