CivArchive
    Wan Video 2.1 - Wan 2.1 VAE

    Wan Video

    Note: There are other Wan Video files hosted on Civitai - these may be duplicates, but this model card is primarily to host the files used by Wan Video in the Civitai Generator.

    👍 SOTA Performance: Wan2.1 consistently outperforms existing open-source models and state-of-the-art commercial solutions across multiple benchmarks.

    • 👍 Supports Consumer-grade GPUs: The T2V-1.3B model requires only 8.19 GB VRAM, making it compatible with almost all consumer-grade GPUs. It can generate a 5-second 480P video on an RTX 4090 in about 4 minutes (without optimization techniques like quantization). Its performance is even comparable to some closed-source models.

    • 👍 Multiple Tasks: Wan2.1 excels in Text-to-Video, Image-to-Video, Video Editing, Text-to-Image, and Video-to-Audio, advancing the field of video generation.

    • 👍 Visual Text Generation: Wan2.1 is the first video model capable of generating both Chinese and English text, featuring robust text generation that enhances its practical applications.

    • 👍 Powerful Video VAE: Wan-VAE delivers exceptional efficiency and performance, encoding and decoding 1080P videos of any length while preserving temporal information, making it an ideal foundation for video and image generation.


    Original HuggingFace Repo: https://huggingface.co/Wan-AI

    Description

    FAQ

    Comments (15)

    NanashiAnonMar 6, 2025· 8 reactions
    CivitAI

    👀

    Interesting.

    EepolMar 7, 2025· 8 reactions
    CivitAI

    Can make with this model erotic content?

    2770379Mar 7, 2025· 3 reactions

    It works "ok" on i2v, but struggles on t2v

    EepolMar 7, 2025· 2 reactions

    @Suppressor thank you

    2770379Mar 7, 2025· 2 reactions

    @Eepol My current opinion lies with Hunyuan for NSFW. That may change as things develop.

    AgkistrodonMar 19, 2025· 1 reaction

    I've made erotic content with it, it's not difficult but it requires NSFW LoRA files. You can see a few of my examples on my profile.

    J1BMar 11, 2025· 5 reactions
    CivitAI

    Can you use this 720P model on a RTX 3090? I just seem to get an out of memory errors (the 480P model works fine) what settings are people with "only" 24GB of Vram using?

    J1BMar 12, 2025· 6 reactions

    Oh I figure out how to do it now, I had to drop the number of frames from 81 to 49 which leaves you with 3 seconds of video, but it fits on a 3090/4090. takes around 30 mins without optimisations though!

    unknowntomanMar 20, 2025· 1 reaction

    @J1B seems to be the same with the GGUF Q8. Nice to know the limits.

    wlmsgApr 26, 2025

    @J1B Generally, there is a memory bottleneck, you can set a higher virtual memory, I only have 4060ti16G and can still use 14B i2v 720p 81 frames.

    AiMetatronApr 27, 2025· 1 reaction

    For video models, 24GVRAM consumes too much time (although block swapping is good), modifying 4090 48GB or A100-40GB stolen from cloud GPU centers is the best choice ; )

    pauldenoiserMay 6, 2025

    @METAFILM_Ai what do you mean whit "modifying 4090 48GB"?

    J1BMay 9, 2025

    @pauldenoiser :https://videocardz.com/newz/custom-geforce-rtx-4090-48gb-now-comes-with-water-cooling-sales-of-modded-48gb-cards-booming-in-china

    nikolatesla20145May 26, 2025· 1 reaction

    @unknowntoman I can easily do 65 frames with the 720p Q8_0 GGUF...

    mmdd2543Jul 28, 2025· 1 reaction

    Check out block swapping. If you are using Kijai's Wan wrapper nodes, it comes with a block swapping node. If you are using a native/standard ComfyUI workflow, you can try to use the ComfyUI-wanBlockswap ComfyUI node by orssorbit.

    Checkpoint
    Wan Video
    by CivitaiOfficial
    Download (Beta) View on CivitAI
    base model
    video generation

    Details

    Downloads
    2,533
    Platform
    CivitAI
    Platform Status
    Available
    Created
    3/6/2025
    Updated
    4/30/2026
    Deleted
    -

    Files

    wanVideo21_wan21VAE.safetensors

    Size:
    242.06 MB
    SHA256:
    2fc39d31359a4b0a64f55876d8ff7fa8d780956ae2cb13463b0223e15148976b

    Mirrors

    Huggingface (113 mirrors)
    CivitAI (4 mirrors)
    TensorFiles (1 mirrors)
    Torrents (1 mirrors)
    Other Platforms (TensorArt, SeaArt, etc.) (5 mirrors)