Wan2.2-TI2V-5B-Diffusers

Wan2.2-TI2V-5B-Diffusers

Makes video clips from a text prompt or an image plus text, at up to 720P and 24 frames per second, in the Diffusers format for a single consumer GPU.

Made by Wan-AI on Hugging Face. These notes are Mindket's, written from its model page and model card; the model's own files and license are on Hugging Face.

About

What it is: Wan2.2-TI2V-5B-Diffusers is the 5B text-image-to-video model of the Wan2.2 family, packaged for the Diffusers library. The same model takes a written prompt alone, or a prompt with a starting image.

It uses the Wan2.2-VAE, which compresses video by 16 by 16 by 4. The card says this lets it produce 720P video at 24 frames per second on one consumer GPU such as an RTX 4090.

Who it suits: Video creators, developers and researchers who want an open video generator that fits on a single high-end gaming GPU instead of a multi-GPU server.

Industry | Who uses it | How they use it | Example
Media | Video creators | Make short clips from a prompt or a still image | Animate a still picture with a text description
Software | Developers | Add video generation to a Python pipeline through Diffusers | Call the model from your own script
Research | Academic labs | Study an efficient 720P video model on one GPU | Compare it with the larger A14B models

People without a capable GPU can skip running it locally.

Where to use it:

• Python with the Diffusers library, since this is the Diffusers version
• ComfyUI, which the card lists as integrated for the 5B model
• Windows, macOS or Linux with Python, for downloading the files
• A consumer GPU such as an RTX 4090, which the card names as enough

Use case | Who | What they get back
Text to video | Creators | A clip made from a written description
Image to video | Creators | A clip that starts from your image and moves
Pipeline integration | Developers | Video generation inside your own Python code

How to set it up: You need Python, a recent torch and a strong GPU.

• Python and pip
• torch 2.4.0 or newer, as the card's install step notes
• A single consumer GPU such as an RTX 4090
• Disk space for the weights, about 20 GB in F32 by Hugging Face's count

1. Clone the Wan2.2 repository from GitHub and move into its folder.
2. Install the requirements with pip.
3. Install the huggingface_hub package.
4. Download the model files with the hf command.

cd Wan2.2
# Ensure torch >= 2.4.0
pip install -r requirements.txt

pip install -U huggingface_hub
hf download Wan-AI/Wan2.2-TI2V-5B-Diffusers

How to use it: Pass a text prompt to get a video, or an image plus a prompt to animate it. Because this is the Diffusers version, it is meant to load with the diffusers library. The card's own run commands and model download table are cut off here, so read them on the model page.

Example: You describe a short scene in words, or give a still image with a few words about the motion you want. The model returns a video, which the card says can reach 720P at 24 frames per second.

For newcomers and experts: Think of it as a video printer that works from a sentence or a picture. A good first try is a short text prompt, since that needs no input image.

Underneath, the Wan2.2 family adds a mixture-of-experts design to video diffusion in its larger models, while this 5B model relies on the Wan2.2-VAE for its compression ratio. The trade-off is a smaller model that runs on one GPU in exchange for the capacity of the A14B models.

Write a prompt or add an image
The Wan2.2-VAE works on compressed video
The 5B diffusion model generates frames
You get a 720P clip

Good to know: The card compares Wan2.2's training data with Wan2.1, which is the maker's own report.

Title: Extra training data versus Wan2.1
Images: 65.6%
Videos: 83.2%

• Size: 5B parameters in Safetensors, about 20 GB in F32 and about 2.5 GB at 4-bit as an estimate
• Languages: English and Chinese are listed
• Variants: Hugging Face lists 1 quantized and 34 fine-tuned versions made by other people
• License: Apache 2.0
• Cost to run: no API key or paid service is named, but you need a strong GPU
• Data it reaches: the prompts and images you give it, plus the files you download from Hugging Face
• Not for: the card names no restrictions beyond its license

On Hugging Face, 13 November 2026

• Task: text to video
• Library: diffusers
• Parameters: 5B
• Weights: about 20 GB in F32; about 2.5 GB at 4-bit (estimate)
• Formats: Safetensors
• Downloads in the last 30 days: 179,371
• Likes: 174
• License: apache-2.0
• Languages: en, zh
• Quantized versions: 1
• Fine-tunes: 35
• Tags: diffusers, safetensors, text-to-video

Links

• Model: https://huggingface.co/Wan-AI/Wan2.2-TI2V-5B-Diffusers
• Files: https://huggingface.co/Wan-AI/Wan2.2-TI2V-5B-Diffusers/tree/main

Get the model

Download its files with Hugging Face's command line:

pip install -U huggingface_hub
hf download Wan-AI/Wan2.2-TI2V-5B-Diffusers

Then follow the usage in its model card. Nothing in this kit runs by itself.