Instructions to use theokiGuyOKI/MiniCPM-V-4.7-1B-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 theokiGuyOKI/MiniCPM-V-4.7-1B-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 theokiGuyOKI/MiniCPM-V-4.7-1B-GGUF:Q4_K_M
# Run inference directly in the terminal:
llama cli -hf theokiGuyOKI/MiniCPM-V-4.7-1B-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 theokiGuyOKI/MiniCPM-V-4.7-1B-GGUF:Q4_K_M
# Run inference directly in the terminal:
llama cli -hf theokiGuyOKI/MiniCPM-V-4.7-1B-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 theokiGuyOKI/MiniCPM-V-4.7-1B-GGUF:Q4_K_M
# Run inference directly in the terminal:
./llama-cli -hf theokiGuyOKI/MiniCPM-V-4.7-1B-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 theokiGuyOKI/MiniCPM-V-4.7-1B-GGUF:Q4_K_M
# Run inference directly in the terminal:
./build/bin/llama-cli -hf theokiGuyOKI/MiniCPM-V-4.7-1B-GGUF:Q4_K_M

Use Docker

docker model run hf.co/theokiGuyOKI/MiniCPM-V-4.7-1B-GGUF:Q4_K_M

LM Studio
Jan

vLLM

How to use theokiGuyOKI/MiniCPM-V-4.7-1B-GGUF with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "theokiGuyOKI/MiniCPM-V-4.7-1B-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": "theokiGuyOKI/MiniCPM-V-4.7-1B-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/theokiGuyOKI/MiniCPM-V-4.7-1B-GGUF:Q4_K_M

Ollama
How to use theokiGuyOKI/MiniCPM-V-4.7-1B-GGUF with Ollama:
```
ollama run hf.co/theokiGuyOKI/MiniCPM-V-4.7-1B-GGUF:Q4_K_M
```
Unsloth Desktop

How to use theokiGuyOKI/MiniCPM-V-4.7-1B-GGUF with Pi:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf theokiGuyOKI/MiniCPM-V-4.7-1B-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": "theokiGuyOKI/MiniCPM-V-4.7-1B-GGUF:Q4_K_M"
        }
      ]
    }
  }
}

Run Pi

# Start Pi in your project directory:
pi

Docker Model Runner
How to use theokiGuyOKI/MiniCPM-V-4.7-1B-GGUF with Docker Model Runner:
```
docker model run hf.co/theokiGuyOKI/MiniCPM-V-4.7-1B-GGUF:Q4_K_M
```

Lemonade

How to use theokiGuyOKI/MiniCPM-V-4.7-1B-GGUF with Lemonade:

Pull the model

# Download Lemonade from https://lemonade-server.ai/
lemonade pull theokiGuyOKI/MiniCPM-V-4.7-1B-GGUF:Q4_K_M

Run and chat with the model

lemonade run user.MiniCPM-V-4.7-1B-GGUF-Q4_K_M

List all available models

lemonade list

Hermes Agent

How to use theokiGuyOKI/MiniCPM-V-4.7-1B-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 theokiGuyOKI/MiniCPM-V-4.7-1B-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 theokiGuyOKI/MiniCPM-V-4.7-1B-GGUF:Q4_K_M

Run Hermes

hermes

Atomic Chat

OpenClaw

How to use theokiGuyOKI/MiniCPM-V-4.7-1B-GGUF with OpenClaw:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf theokiGuyOKI/MiniCPM-V-4.7-1B-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 "theokiGuyOKI/MiniCPM-V-4.7-1B-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"

MiniCPM-V-4.7-1B — GGUF

Community GGUF conversions of openbmb/MiniCPM-V-4.7-1B for llama.cpp multimodal inference.

Official OpenBMB GGUFs: openbmb/MiniCPM-V-4.7-1B-gguf.
This repo adds a full low-to-high quant ladder (Q2_K → Q8_0 + F16) with both 16× and 4× vision projectors, converted with a verified current llama.cpp commit.

Quick start

Download one language model + one mmproj, then run:

llama-mtmd-cli \
  -m MiniCPM-V-4_7-1B-Q4_K_M.gguf \
  --mmproj mmproj-model-f16.gguf \
  -c 8192 \
  --image photo.jpg \
  -p "What is in this image?"

Server example:

llama-server \
  -m MiniCPM-V-4_7-1B-Q4_K_M.gguf \
  --mmproj mmproj-model-f16.gguf \
  --port 8080

Vision always needs an mmproj. Pair any LM quant with mmproj-model-f16.gguf (default 16×) or mmproj-model-f16-4x.gguf (finer detail, more vision tokens).

Which quant should I pick?

Goal	Use
Best balance (recommended)	`Q4_K_M`
Better quality	`Q5_K_M` or `Q6_K`
Near-lossless	`Q8_0`
Max fidelity / re-quantize source	`F16`
Low RAM	`Q3_K_M`
Extremely low RAM	`Q2_K` (expect quality loss)

Files

Language model

File	Size	Notes
`MiniCPM-V-4_7-1B-Q2_K.gguf`	482 MB	Smallest; quality tradeoff
`MiniCPM-V-4_7-1B-Q3_K_S.gguf`	519 MB	Low RAM
`MiniCPM-V-4_7-1B-Q3_K_M.gguf`	549 MB	Low RAM (better than S)
`MiniCPM-V-4_7-1B-Q4_0.gguf`	614 MB	Legacy Q4
`MiniCPM-V-4_7-1B-Q4_1.gguf`	659 MB	Legacy Q4
`MiniCPM-V-4_7-1B-Q4_K_S.gguf`	618 MB	Compact K-quant
`MiniCPM-V-4_7-1B-Q4_K_M.gguf`	641 MB	Recommended default
`MiniCPM-V-4_7-1B-Q5_0.gguf`	704 MB	Legacy Q5
`MiniCPM-V-4_7-1B-Q5_1.gguf`	749 MB	Legacy Q5
`MiniCPM-V-4_7-1B-Q5_K_S.gguf`	704 MB	Higher quality K-quant
`MiniCPM-V-4_7-1B-Q5_K_M.gguf`	718 MB	High quality default alternative
`MiniCPM-V-4_7-1B-Q6_K.gguf`	799 MB	Very high quality
`MiniCPM-V-4_7-1B-Q8_0.gguf`	1031 MB	Near-lossless
`MiniCPM-V-4_7-1B-F16.gguf`	1931 MB	Full precision GGUF

Vision projector (required for images)

File	Size	Notes
`mmproj-model-f16.gguf`	1059 MB	Default 16× compression
`mmproj-model-f16-4x.gguf`	859 MB	4× mode (more visual detail)

mmproj files are kept at F16 on purpose; only the language model is quantized.

Model summary


Base model	`openbmb/MiniCPM-V-4.7-1B`
Architecture	`MiniCPMV4_7ForConditionalGeneration`
Task	Image-text-to-text (multimodal)
License	Apache-2.0

MiniCPM-V 4.7 1B is a compact multimodal model from OpenBMB with strong efficiency for on-device / local image and video understanding. See the upstream model card for benchmarks and Transformers usage.

Conversion details


Toolchain	`llama.cpp` `@4413a19` (`4413a192410fad1f054e9216114087a3e01527a0`)
Method	`convert_hf_to_gguf.py` → `llama-quantize`
Guide	MiniCPM-V 4.7 GGUF cookbook
Note	Requires MiniCPM-V 4.7 support (canvas M-RoPE / PR `#29416`+)

# Language model (F16)
python convert_hf_to_gguf.py ./MiniCPM-V-4.7-1B \
  --outfile MiniCPM-V-4_7-1B-F16.gguf --outtype f16

# Vision projector (16x)
python convert_hf_to_gguf.py ./MiniCPM-V-4.7-1B \
  --mmproj --outfile mmproj-model-f16.gguf

# Quantize LM only
llama-quantize MiniCPM-V-4_7-1B-F16.gguf MiniCPM-V-4_7-1B-Q4_K_M.gguf Q4_K_M

Intended use

Local / on-device multimodal chat with llama.cpp
Offline demos and edge experiments where a small VLM is enough

Limitations

Not a drop-in replacement for the original BF16 checkpoint for training or exact numeric reproduction
Lower-bit quants (Q2_K / Q3_*) can hurt OCR, grounding, and long reasoning
Outputs may be inaccurate or biased; keep human review for important decisions
Use a recent llama.cpp build that supports MiniCPM-V 4.7

License & credit

Derived from MiniCPM-V 4.7 1B under Apache-2.0
All credit for the original model to OpenBMB / ModelBest
Upstream: https://github.com/OpenBMB/MiniCPM-o
Please cite / star the upstream project if you use this work

@article{yao2024minicpm,
  title={MiniCPM-V: A GPT-4V Level MLLM on Your Phone},
  author={Yao, Yuan and Yu, Tianyu and Zhang, Ao and Wang, Chongyi and Cui, Junbo and Zhu, Hongji and Cai, Tianchi and Li, Haoyu and Zhao, Weilin and He, Zhihui and others},
  journal={arXiv preprint arXiv:2408.01800},
  year={2024}
}

Downloads last month: 864

GGUF

Model size

1B params

Architecture

qwen35

Hardware compatibility

2-bit

3-bit

4-bit

5-bit

6-bit

8-bit

16-bit

Model tree for theokiGuyOKI/MiniCPM-V-4.7-1B-GGUF

Base model

openbmb/MiniCPM-V-4.7-1B

Quantized

(49)

this model

Paper for theokiGuyOKI/MiniCPM-V-4.7-1B-GGUF

MiniCPM-V: A GPT-4V Level MLLM on Your Phone

Paper • 2408.01800 • Published Aug 3, 2024 • 95