Skip to content
SPECIFICTION
Get listed

Wan

github.com

LinuxSelf-hosted
Visit github.com
Spec score 7.4 #184 of 242 in AI video generators

What it is

Wan is a free, open-source suite of video foundation models that runs on Linux and can be self-hosted. It supports text-to-video, image-to-video, video editing, text-to-image, and video-to-audio generation, including Chinese and English text inside videos. VACE handles editing or generation from a prompt with optional video, mask, and image inputs. T2V-14B supports 480P and 720P output; T2V-1.3B supports 480P, while its 720P results are described as less stable. The T2V-1.3B model requires 8.19 GB of VRAM, and the project reports about four minutes for a five-second 480P video on an RTX 4090 without quantization. Wan-VAE encodes and decodes 1080P video of any length. Model weights are available through Hugging Face and ModelScope, with Diffusers and ComfyUI integrations. Installation uses pip or Poetry and requires PyTorch 2.4.0 or later. The Apache-2.0 license permits commercial use.

Spec sheet

Product
Wan2.1 is an open suite of video foundation models for video generation.Read 8 Oct 2026 · github.com
Integrations
The project says Wan2.1 T2V and I2V were integrated into Diffusers and that Wan2.1 was integrated into ComfyUI.Read 8 Oct 2026 · github.com
Support
The repository directs users to Discord and WeChat groups to contact the research or product teams.Read 8 Oct 2026 · github.com
License
The repository identifies its license as Apache-2.0.Read 8 Oct 2026 · github.com
Commercial use
YesRead 19 Sep 2026 · github.com
Downloads
Model weights are linked for download through Hugging Face and ModelScope.Read 8 Oct 2026 · github.com
Generation tasks
The models support text-to-video, image-to-video, video editing, text-to-image, and video-to-audio tasks.Read 8 Oct 2026 · github.com
GPU requirement
The T2V-1.3B model requires 8.19 GB VRAM according to the repository, which also reports about four minutes to generate a five-second 480P video on an RTX 4090 without quantization.Read 8 Oct 2026 · github.com
Hardware requirement
The T2V-1.3B model requires 8.19 GB of VRAM, and the project says it can generate a five-second 480P video on an RTX 4090 in about four minutes without quantization.Read 8 Oct 2026 · github.com
Image and video tasks
The repository lists image-to-video, video editing, text-to-image, and video-to-audio among its supported tasks.Read 8 Oct 2026 · github.com
Image-to-video
YesRead 19 Sep 2026 · github.com
Installation
The installation guide documents pip and Poetry installation methods.Read 8 Oct 2026 · github.com
Local interface
The repository includes instructions for running a local Gradio interface.Read 8 Oct 2026 · github.com
Maximum resolution
720pRead 19 Sep 2026 · github.com
Maximum video length
0.083333Read 19 Sep 2026 · github.com
Model downloads
The project provides model downloads through Hugging Face and ModelScope.Read 8 Oct 2026 · github.com
Prompt extension
Prompt extension can use the DashScope API or a local Qwen model.Read 8 Oct 2026 · github.com
Requirements
The quickstart says to ensure PyTorch 2.4.0 or later before installing dependencies.Read 8 Oct 2026 · github.com
Resolution support
The T2V-14B model supports 480P and 720P, while T2V-1.3B supports 480P.Read 8 Oct 2026 · github.com
Resolutions
The repository lists T2V-14B at 480P and 720P, and T2V-1.3B at 480P; it cautions that 1.3B results at 720P are less stable than at 480P.Read 8 Oct 2026 · github.com
Runtime
The quickstart instructs users to install dependencies with PyTorch 2.4.0 or later and provides command-line and local Gradio inference examples.Read 8 Oct 2026 · github.com
Text generation
Wan2.1 can generate Chinese and English text within videos.Read 8 Oct 2026 · github.com
Text to video
Wan2.1 supports text-to-video generation with 1.3B and 14B models.Read 8 Oct 2026 · github.com
Video editing
VACE accepts a text prompt and optional video, mask, and image inputs for video generation or editing.Read 8 Oct 2026 · github.com
Video VAE
Wan-VAE encodes and decodes 1080P videos of any length while preserving temporal information.Read 8 Oct 2026 · github.com
Visual text
The maker says the model can generate Chinese and English text in video.Read 8 Oct 2026 · github.com

Strong points and limits

Strong points

  • Supports text-to-video and image-to-video generation
  • Includes video editing, text-to-image, and video-to-audio
  • Apache-2.0 license permits commercial use
  • Runs locally on Linux with self-hosting
  • Model weights available from Hugging Face and ModelScope

Limits

  • T2V-1.3B requires 8.19 GB of VRAM
  • Maximum listed resolution is 720P
  • T2V-1.3B 720P results are less stable
  • Requires PyTorch 2.4.0 or later

Who it suits

Wan suits developers and teams wanting self-hosted video generation on Linux. It is also relevant to users who need commercial-use rights and integrations with Diffusers or ComfyUI.

Verdict

Wan provides a broad, locally deployable video-model suite with commercial-use licensing. Hardware and Linux setup requirements matter, especially for higher-resolution generation.

Where Specifiction places it

Alternatives to WanAll alternatives →