Instructions to use prithivMLmods/ProactiveMobile-Qwen2.5-VL-7B-GGUF with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use prithivMLmods/ProactiveMobile-Qwen2.5-VL-7B-GGUF with Transformers:

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

pipe = pipeline("image-text-to-text", model="prithivMLmods/ProactiveMobile-Qwen2.5-VL-7B-GGUF")

# pip install -U transformers accelerate
# Load model directly
from transformers import AutoModel
model = AutoModel.from_pretrained("prithivMLmods/ProactiveMobile-Qwen2.5-VL-7B-GGUF", device_map="auto")

Notebooks
Google Colab
Kaggle
Local Apps Settings

llama.cpp

How to use prithivMLmods/ProactiveMobile-Qwen2.5-VL-7B-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 prithivMLmods/ProactiveMobile-Qwen2.5-VL-7B-GGUF:Q4_K_M
# Run inference directly in the terminal:
llama cli -hf prithivMLmods/ProactiveMobile-Qwen2.5-VL-7B-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 prithivMLmods/ProactiveMobile-Qwen2.5-VL-7B-GGUF:Q4_K_M
# Run inference directly in the terminal:
llama cli -hf prithivMLmods/ProactiveMobile-Qwen2.5-VL-7B-GGUF:Q4_K_M

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 prithivMLmods/ProactiveMobile-Qwen2.5-VL-7B-GGUF:Q4_K_M
# Run inference directly in the terminal:
./build/bin/llama cli -hf prithivMLmods/ProactiveMobile-Qwen2.5-VL-7B-GGUF:Q4_K_M

Use Docker

docker model run hf.co/prithivMLmods/ProactiveMobile-Qwen2.5-VL-7B-GGUF:Q4_K_M

Llama new
LM Studio
Jan

vLLM

How to use prithivMLmods/ProactiveMobile-Qwen2.5-VL-7B-GGUF with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "prithivMLmods/ProactiveMobile-Qwen2.5-VL-7B-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": "prithivMLmods/ProactiveMobile-Qwen2.5-VL-7B-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/prithivMLmods/ProactiveMobile-Qwen2.5-VL-7B-GGUF:Q4_K_M

SGLang

How to use prithivMLmods/ProactiveMobile-Qwen2.5-VL-7B-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 "prithivMLmods/ProactiveMobile-Qwen2.5-VL-7B-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": "prithivMLmods/ProactiveMobile-Qwen2.5-VL-7B-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 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 "prithivMLmods/ProactiveMobile-Qwen2.5-VL-7B-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": "prithivMLmods/ProactiveMobile-Qwen2.5-VL-7B-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"
						}
					}
				]
			}
		]
	}'

Ollama
How to use prithivMLmods/ProactiveMobile-Qwen2.5-VL-7B-GGUF with Ollama:
```
ollama run hf.co/prithivMLmods/ProactiveMobile-Qwen2.5-VL-7B-GGUF:Q4_K_M
```
Unsloth Desktop
Docker Model Runner
How to use prithivMLmods/ProactiveMobile-Qwen2.5-VL-7B-GGUF with Docker Model Runner:
```
docker model run hf.co/prithivMLmods/ProactiveMobile-Qwen2.5-VL-7B-GGUF:Q4_K_M
```

Lemonade

How to use prithivMLmods/ProactiveMobile-Qwen2.5-VL-7B-GGUF with Lemonade:

Pull the model

# Download Lemonade from https://lemonade-server.ai/
lemonade pull prithivMLmods/ProactiveMobile-Qwen2.5-VL-7B-GGUF:Q4_K_M

Run and chat with the model

lemonade run user.ProactiveMobile-Qwen2.5-VL-7B-GGUF-Q4_K_M

List all available models

lemonade list

Atomic Chat

ProactiveMobile-Qwen2.5-VL-7B-GGUF

ProactiveMobile Agent (xiaomi-research/ProactiveMobile-Qwen2.5-VL-7B) is a 7B multimodal proactive mobile-assistant model from Xiaomi Research, introduced at CVPR 2026 and fine-tuned from Qwen/Qwen2.5-VL-7B-Instruct on the Chinese-language ProactiveMobile dataset. Given user context and GUI screenshots or textual action traces, it predicts whether to intervene or remain silent, producing structured reasoning, recommendations, and function calls. Across 3,658 benchmark instances, it achieves 49.29% Type-Acc, 54.23% precision, 54.19% recall, 53.83% F1, and an 8.86% False Trigger Rate (FTR). Results vary by difficulty: L1 achieves 72.41% Type-Acc and 75.16% F1, L2 57.93% and 63.10%, and L3 38.99% and 43.45%. Text traces perform better, reaching 64.04% Type-Acc, 65.46% F1, and 1.31% FTR, compared with 34.57%, 42.23%, and 24.37% for GUI traces. The weights use Apache-2.0, while the dataset carries CC BY-NC-SA 4.0 non-commercial share-alike restrictions. Supports Hugging Face Transformers via Qwen2_5_VLForConditionalGeneration and AutoProcessor.

Model Files

File Name	Quant Type	File Size	File Link	Description
ProactiveMobile-Qwen2.5-VL-7B.BF16.gguf	BF16	15.2 GB	Link	Full BF16 weights. Highest quality, largest file size.
ProactiveMobile-Qwen2.5-VL-7B.Q3_K_L.gguf	Q3_K_L	4.09 GB	Link	Lower quality but usable, good for low RAM availability.
ProactiveMobile-Qwen2.5-VL-7B.Q3_K_M.gguf	Q3_K_M	3.81 GB	Link	Low quality.
ProactiveMobile-Qwen2.5-VL-7B.Q4_K_M.gguf	Q4_K_M	4.68 GB	Link	Good quality, default size for most use cases, recommended.
ProactiveMobile-Qwen2.5-VL-7B.Q4_K_S.gguf	Q4_K_S	4.46 GB	Link	Slightly lower quality with more space savings, recommended.
ProactiveMobile-Qwen2.5-VL-7B.Q5_K_M.gguf	Q5_K_M	5.44 GB	Link	High quality, recommended.
ProactiveMobile-Qwen2.5-VL-7B.Q5_K_S.gguf	Q5_K_S	5.32 GB	Link	High quality, recommended.
ProactiveMobile-Qwen2.5-VL-7B.Q6_K.gguf	Q6_K	6.25 GB	Link	Very high quality, near perfect, recommended.
ProactiveMobile-Qwen2.5-VL-7B.Q8_0.gguf	Q8_0	8.1 GB	Link	Extremely high quality, generally unneeded but max available quant.
ProactiveMobile-Qwen2.5-VL-7B.mmproj-bf16.gguf	mmproj-bf16	1.36 GB	Link	Multimodal projection file in BF16 format. Used for vision/language models.