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 Settings
- 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
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
- Atomic Chat new
- 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
File size: 5,613 Bytes
e00eceb | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 | from typing import Optional
import torch
from pydantic import BaseModel, Field
from typing_extensions import override
from comfy_api.latest import IO, ComfyExtension
from comfy_api_nodes.util import (
ApiEndpoint,
download_url_to_video_output,
get_number_of_images,
poll_op,
sync_op,
tensor_to_bytesio,
)
class Sora2GenerationRequest(BaseModel):
prompt: str = Field(...)
model: str = Field(...)
seconds: str = Field(...)
size: str = Field(...)
class Sora2GenerationResponse(BaseModel):
id: str = Field(...)
error: Optional[dict] = Field(None)
status: Optional[str] = Field(None)
class OpenAIVideoSora2(IO.ComfyNode):
@classmethod
def define_schema(cls):
return IO.Schema(
node_id="OpenAIVideoSora2",
display_name="OpenAI Sora - Video",
category="api node/video/Sora",
description="OpenAI video and audio generation.",
inputs=[
IO.Combo.Input(
"model",
options=["sora-2", "sora-2-pro"],
default="sora-2",
),
IO.String.Input(
"prompt",
multiline=True,
default="",
tooltip="Guiding text; may be empty if an input image is present.",
),
IO.Combo.Input(
"size",
options=[
"720x1280",
"1280x720",
"1024x1792",
"1792x1024",
],
default="1280x720",
),
IO.Combo.Input(
"duration",
options=[4, 8, 12],
default=8,
),
IO.Image.Input(
"image",
optional=True,
),
IO.Int.Input(
"seed",
default=0,
min=0,
max=2147483647,
step=1,
display_mode=IO.NumberDisplay.number,
control_after_generate=True,
optional=True,
tooltip="Seed to determine if node should re-run; "
"actual results are nondeterministic regardless of seed.",
),
],
outputs=[
IO.Video.Output(),
],
hidden=[
IO.Hidden.auth_token_comfy_org,
IO.Hidden.api_key_comfy_org,
IO.Hidden.unique_id,
],
is_api_node=True,
price_badge=IO.PriceBadge(
depends_on=IO.PriceBadgeDepends(widgets=["model", "size", "duration"]),
expr="""
(
$m := widgets.model;
$size := widgets.size;
$dur := widgets.duration;
$isPro := $contains($m, "sora-2-pro");
$isSora2 := $contains($m, "sora-2");
$isProSize := ($size = "1024x1792" or $size = "1792x1024");
$perSec :=
$isPro ? ($isProSize ? 0.5 : 0.3) :
$isSora2 ? 0.1 :
($isProSize ? 0.5 : 0.1);
{"type":"usd","usd": $round($perSec * $dur, 2)}
)
""",
),
)
@classmethod
async def execute(
cls,
model: str,
prompt: str,
size: str = "1280x720",
duration: int = 8,
seed: int = 0,
image: Optional[torch.Tensor] = None,
):
if model == "sora-2" and size not in ("720x1280", "1280x720"):
raise ValueError("Invalid size for sora-2 model, only 720x1280 and 1280x720 are supported.")
files_input = None
if image is not None:
if get_number_of_images(image) != 1:
raise ValueError("Currently only one input image is supported.")
files_input = {"input_reference": ("image.png", tensor_to_bytesio(image), "image/png")}
initial_response = await sync_op(
cls,
endpoint=ApiEndpoint(path="/proxy/openai/v1/videos", method="POST"),
data=Sora2GenerationRequest(
model=model,
prompt=prompt,
seconds=str(duration),
size=size,
),
files=files_input,
response_model=Sora2GenerationResponse,
content_type="multipart/form-data",
)
if initial_response.error:
raise Exception(initial_response.error["message"])
model_time_multiplier = 1 if model == "sora-2" else 2
await poll_op(
cls,
poll_endpoint=ApiEndpoint(path=f"/proxy/openai/v1/videos/{initial_response.id}"),
response_model=Sora2GenerationResponse,
status_extractor=lambda x: x.status,
poll_interval=8.0,
estimated_duration=int(45 * (duration / 4) * model_time_multiplier),
)
return IO.NodeOutput(
await download_url_to_video_output(f"/proxy/openai/v1/videos/{initial_response.id}/content", cls=cls),
)
class OpenAISoraExtension(ComfyExtension):
@override
async def get_node_list(self) -> list[type[IO.ComfyNode]]:
return [
OpenAIVideoSora2,
]
async def comfy_entrypoint() -> OpenAISoraExtension:
return OpenAISoraExtension()
|