Instructions to use vidfom/Ltx-3 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- llama-cpp-python
How to use vidfom/Ltx-3 with llama-cpp-python:
# !pip install llama-cpp-python from llama_cpp import Llama llm = Llama.from_pretrained( repo_id="vidfom/Ltx-3", filename="ComfyUI/models/text_encoders/gemma-3-12b-it-qat-UD-Q4_K_XL.gguf", )
llm.create_chat_completion( messages = "No input example has been defined for this model task." )
- Notebooks
- Google Colab
- Kaggle
- Local Apps
- llama.cpp
How to use vidfom/Ltx-3 with llama.cpp:
Install from brew
brew install llama.cpp # Start a local OpenAI-compatible server with a web UI: llama-server -hf vidfom/Ltx-3:UD-Q4_K_XL # Run inference directly in the terminal: llama-cli -hf vidfom/Ltx-3:UD-Q4_K_XL
Install from WinGet (Windows)
winget install llama.cpp # Start a local OpenAI-compatible server with a web UI: llama-server -hf vidfom/Ltx-3:UD-Q4_K_XL # Run inference directly in the terminal: llama-cli -hf vidfom/Ltx-3:UD-Q4_K_XL
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 vidfom/Ltx-3:UD-Q4_K_XL # Run inference directly in the terminal: ./llama-cli -hf vidfom/Ltx-3:UD-Q4_K_XL
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 vidfom/Ltx-3:UD-Q4_K_XL # Run inference directly in the terminal: ./build/bin/llama-cli -hf vidfom/Ltx-3:UD-Q4_K_XL
Use Docker
docker model run hf.co/vidfom/Ltx-3:UD-Q4_K_XL
- LM Studio
- Jan
- Ollama
How to use vidfom/Ltx-3 with Ollama:
ollama run hf.co/vidfom/Ltx-3:UD-Q4_K_XL
- Unsloth Studio new
How to use vidfom/Ltx-3 with Unsloth Studio:
Install Unsloth Studio (macOS, Linux, WSL)
curl -fsSL https://unsloth.ai/install.sh | sh # Run unsloth studio unsloth studio -H 0.0.0.0 -p 8888 # Then open http://localhost:8888 in your browser # Search for vidfom/Ltx-3 to start chatting
Install Unsloth Studio (Windows)
irm https://unsloth.ai/install.ps1 | iex # Run unsloth studio unsloth studio -H 0.0.0.0 -p 8888 # Then open http://localhost:8888 in your browser # Search for vidfom/Ltx-3 to start chatting
Using HuggingFace Spaces for Unsloth
# No setup required # Open https://huggingface.co/spaces/unsloth/studio in your browser # Search for vidfom/Ltx-3 to start chatting
- Docker Model Runner
How to use vidfom/Ltx-3 with Docker Model Runner:
docker model run hf.co/vidfom/Ltx-3:UD-Q4_K_XL
- Lemonade
How to use vidfom/Ltx-3 with Lemonade:
Pull the model
# Download Lemonade from https://lemonade-server.ai/ lemonade pull vidfom/Ltx-3:UD-Q4_K_XL
Run and chat with the model
lemonade run user.Ltx-3-UD-Q4_K_XL
List all available models
lemonade list
| import pytest | |
| import base64 | |
| import json | |
| import struct | |
| from io import BytesIO | |
| from PIL import Image | |
| from aiohttp import web | |
| from unittest.mock import patch | |
| from app.model_manager import ModelFileManager | |
| pytestmark = ( | |
| pytest.mark.asyncio | |
| ) # This applies the asyncio mark to all test functions in the module | |
| def model_manager(): | |
| return ModelFileManager() | |
| def app(model_manager): | |
| app = web.Application() | |
| routes = web.RouteTableDef() | |
| model_manager.add_routes(routes) | |
| app.add_routes(routes) | |
| return app | |
| async def test_get_model_preview_safetensors(aiohttp_client, app, tmp_path): | |
| img = Image.new('RGB', (100, 100), 'white') | |
| img_byte_arr = BytesIO() | |
| img.save(img_byte_arr, format='PNG') | |
| img_byte_arr.seek(0) | |
| img_b64 = base64.b64encode(img_byte_arr.getvalue()).decode('utf-8') | |
| safetensors_file = tmp_path / "test_model.safetensors" | |
| header_bytes = json.dumps({ | |
| "__metadata__": { | |
| "ssmd_cover_images": json.dumps([img_b64]) | |
| } | |
| }).encode('utf-8') | |
| length_bytes = struct.pack('<Q', len(header_bytes)) | |
| with open(safetensors_file, 'wb') as f: | |
| f.write(length_bytes) | |
| f.write(header_bytes) | |
| with patch('folder_paths.folder_names_and_paths', { | |
| 'test_folder': ([str(tmp_path)], None) | |
| }): | |
| client = await aiohttp_client(app) | |
| response = await client.get('/experiment/models/preview/test_folder/0/test_model.safetensors') | |
| # Verify response | |
| assert response.status == 200 | |
| assert response.content_type == 'image/webp' | |
| # Verify the response contains valid image data | |
| img_bytes = BytesIO(await response.read()) | |
| img = Image.open(img_bytes) | |
| assert img.format | |
| assert img.format.lower() == 'webp' | |
| # Clean up | |
| img.close() | |