0
Video generationApache-2.0Python

FramePack

Long image-to-video generation on laptop GPUs with as little as 6 GB VRAM.

GitHub
Stars
17.3K
Forks
1.7K
Open issues
492
Last commit
2025-10-16

Live from the GitHub API · updated 2026-10-15

What is FramePack?

FramePack, by the author of ControlNet, compresses past frames into a fixed context so video length does not increase memory. It turns a single image into minute-long clips on consumer GPUs, previewing frames as they render.

Key features

  • Image-to-video with progressive frame preview
  • Constant memory regardless of clip length
  • Runs on 6 GB laptop GPUs
  • Windows one-click package

System requirements

Platforms
Windows · Linux
GPU
NVIDIA RTX 30/40/50 series (fp16/bf16 support)
VRAM
6 GB minimum
Disk
~40 GB for models

How to install FramePack

Summarized from the project’s README. Projects change quickly — check the GitHub page for the latest instructions.

Windows one-click package

  1. 1
    Download the one-click package (CUDA + PyTorch included) from the README and extract it.
  2. 2
    Run update.bat, then run.bat. Models download on first launch.

Linux

  1. 1
    Install PyTorch with CUDA, then the requirements.
    git clone https://github.com/lllyasviel/FramePack.git
cd FramePack
pip install -r requirements.txt
  2. 2
    Start the Gradio app.
    python demo_gradio.py

Strengths

  • Very low VRAM requirement
  • Long clips
  • Simple Windows setup

Limitations

  • Image-to-video only
  • Slow per second of video on small GPUs

FramePack alternatives compared

SoftwareStarsLicensePlatformsVRAM
FramePack

Long image-to-video generation on laptop GPUs with as little as 6 GB VRAM.

17.3KApache-2.06 GB minimum
Wan 2.2

Alibaba’s open-weight video generation models (text, image and speech to video).

17.9KApache-2.024 GB for TI2V-5B at 720p; 80 GB-class GPUs (or offloading) for A14B models
LTX-Video

Fast, real-time-capable open video generation from Lightricks.

11.1KApache-2.012 GB+ for smaller variants; more for 13B models
AnimateDiff

Animate your Stable Diffusion checkpoints with plug-in motion modules.

12.3KApache-2.08–12 GB
HunyuanVideo

Tencent’s 13B-parameter open video foundation model.

12.6KTencent Hunyuan Community License45 GB minimum, 60–80 GB recommended for 720p (per README); less with community quantizations
CogVideoX

Open text-to-video and image-to-video models from Zhipu AI / Tsinghua.

13.1KApache-2.0~12–24 GB depending on model and optimizations (see the README table)

Related models

WanAlibaba’s open-weight video model you can run yourself.Kling 2.5 TurboKuaishou’s fast, affordable video model with fluid, stable motion.

FAQ

Is FramePack free?

Yes — FramePack is open-source software released under the Apache-2.0 license. Check the license (and any model weight licenses) before commercial use.

Does FramePack run on Windows, Mac and Linux?

FramePack supports: Windows, Linux.

What hardware do I need for FramePack?

NVIDIA RTX 30/40/50 series (fp16/bf16 support). VRAM: 6 GB minimum

How do I install FramePack?

1. Download the one-click package (CUDA + PyTorch included) from the README and extract it. 2. Run update.bat, then run.bat. Models download on first launch.

What are the best FramePack alternatives?

Popular alternatives include Wan 2.2, LTX-Video, AnimateDiff, HunyuanVideo. See the comparison table on this page.

Can I use FramePack online without installing it?

If you don’t have a large GPU, AiLoft runs Veo 3.1 and Kling 2.5 in the cloud — generate and compare clips in the browser.

No 24 GB GPU? Generate AI video online

Run Veo 3.1 and Kling 2.5 in the cloud and compare clips side by side — pay per clip.

AI Video Generator
Open-source AI video generators