Instructions to use Thomas0202/promptcut-ia with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Notebooks
Google Colab
Kaggle
Local Apps Settings

How to use Thomas0202/promptcut-ia 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 Thomas0202/promptcut-ia:Q8_0
# Run inference directly in the terminal:
llama cli -hf Thomas0202/promptcut-ia:Q8_0

Install from WinGet (Windows)

winget install llama.cpp
# Start a local OpenAI-compatible server with a web UI:
llama serve -hf Thomas0202/promptcut-ia:Q8_0
# Run inference directly in the terminal:
llama cli -hf Thomas0202/promptcut-ia:Q8_0

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 Thomas0202/promptcut-ia:Q8_0
# Run inference directly in the terminal:
./build/bin/llama cli -hf Thomas0202/promptcut-ia:Q8_0

Use Docker

docker model run hf.co/Thomas0202/promptcut-ia:Q8_0

Llama new
LM Studio
Jan

vLLM

How to use Thomas0202/promptcut-ia with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "Thomas0202/promptcut-ia"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "Thomas0202/promptcut-ia",
		"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/Thomas0202/promptcut-ia:Q8_0

Ollama
How to use Thomas0202/promptcut-ia with Ollama:
```
ollama run hf.co/Thomas0202/promptcut-ia:Q8_0
```
Unsloth Desktop

How to use Thomas0202/promptcut-ia with Pi:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf Thomas0202/promptcut-ia:Q8_0

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": "Thomas0202/promptcut-ia:Q8_0"
        }
      ]
    }
  }
}

Run Pi

# Start Pi in your project directory:
pi

Docker Model Runner
How to use Thomas0202/promptcut-ia with Docker Model Runner:
```
docker model run hf.co/Thomas0202/promptcut-ia:Q8_0
```

Lemonade

How to use Thomas0202/promptcut-ia with Lemonade:

Pull the model

# Download Lemonade from https://lemonade-server.ai/
lemonade pull Thomas0202/promptcut-ia:Q8_0

Run and chat with the model

lemonade run user.promptcut-ia-Q8_0

List all available models

lemonade list

Hermes Agent

How to use Thomas0202/promptcut-ia with Hermes Agent:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf Thomas0202/promptcut-ia:Q8_0

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 Thomas0202/promptcut-ia:Q8_0

Run Hermes

hermes

Atomic Chat

OpenClaw

How to use Thomas0202/promptcut-ia with OpenClaw:

Start the llama.cpp server

# Install llama.cpp:
brew install llama.cpp
# Start a local OpenAI-compatible server:
llama serve -hf Thomas0202/promptcut-ia:Q8_0

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 "Thomas0202/promptcut-ia:Q8_0" \
  --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"

PROMPTCUT IA

The on-device brain of PROMPTCUT, a free video editor for iPhone. It watches your clip, reads what you say and decides which graphics to add, where and when — titles, lower thirds, prices, checklists, maps, logos — then checks its own work like a second editor would.

Everything runs on the phone: no account, no server, no internet needed after the download. Your videos never leave your device.


Base model	Qwen3.5-2B (Apache 2.0)
Training	LoRA fine-tune on editing plans and critiques written by a large teacher model, reviewed by hand
Languages	Italian, English, Spanish
Runs on	iPhone (Metal) via llama.cpp / llama.rn
License	CC BY-NC 4.0 — free for personal & research use, no commercial use

What it does

🎬 Director — from the transcript and a few frames it writes a plan: a list of graphics with type, style, text and timing, in a strict JSON format the app can render.
🔍 Critic — it re-reads the plan and flags what doesn't work (text over a face, too many graphics at once, wrong timing) so the app can fix it.
👁️ Vision — it looks at frames to avoid covering faces and to pick colors and positions that read well.
💬 Chat — "add the price at the start", "make it more minimal": it edits the project for you.

Files

File	What	Size
`promptcut-002-Q4_K_M.gguf`	language model, 4-bit	~1.3 GB
`mmproj-promptcut-002-q8_0.gguf`	vision projector, 8-bit	~0.36 GB

The PROMPTCUT app downloads both automatically from Settings → PROMPTCUT IA → Download. You only need them both if you want to try the model elsewhere.

Use with llama.cpp

llama-mtmd-cli -m promptcut-002-Q4_K_M.gguf --mmproj mmproj-promptcut-002-q8_0.gguf \
  --image frame.jpg -p "Plan the graphics for this clip: ..."

The model is tuned for PROMPTCUT's own prompts and output schema: as a general chat model it is just a small Qwen.

License

The PROMPTCUT IA weights (both .gguf files) are released under Creative Commons Attribution-NonCommercial 4.0:

✅ you can download, use, study, modify and share them for personal, research and educational purposes, giving credit to PROMPTCUT IA by Thomas Colombo;
❌ you cannot use them, or anything derived from them, for commercial purposes (paid products, services, ads, client work) without written permission.

The base model Qwen3.5-2B is © Alibaba Cloud under Apache 2.0: see NOTICE. For commercial licensing write to thomas.colombo06@gmail.com.

Limits

It is a 2B model: great at its job, not a general assistant. It can still pick the wrong graphic or misread a frame — the app always lets you change everything before export.

🇮🇹 In italiano

PROMPTCUT IA è il modello che gira dentro PROMPTCUT, editor video gratuito per iPhone: guarda il video, legge il parlato e decide quali grafiche mettere, dove e quando, poi ricontrolla il lavoro. Tutto sul telefono, senza account e senza server. Basato su Qwen3.5-2B (Apache 2.0); i pesi di PROMPTCUT IA sono sotto licenza CC BY-NC 4.0: liberi per uso personale e di ricerca, vietato l'uso commerciale senza permesso scritto.

🇪🇸 En español

PROMPTCUT IA es el modelo que funciona dentro de PROMPTCUT, editor de vídeo gratuito para iPhone: mira el vídeo, lee lo que se dice y decide qué gráficos poner, dónde y cuándo, y luego revisa el trabajo. Todo en el teléfono, sin cuenta y sin servidores. Basado en Qwen3.5-2B (Apache 2.0); los pesos de PROMPTCUT IA tienen licencia CC BY-NC 4.0: libres para uso personal y de investigación, prohibido el uso comercial sin permiso escrito.