CivArchive
    MiniMax H3 pruned fp8 scaled - fl2va pruned fp8 scaled
    Preview 138714647

    MiniMax H3 pruned fp8 scaled

    Pruned FP8 version of the video model

    fl2va:

    (I2V, FL2VA, T2V)

    Direct link: https://huggingface.co/Comfy-Org/MiniMax-H3/resolve/main/diffusion_models/minimax_h3_fl2va_pruned_fp8_scaled.safetensors?download=true

    ref2va:

    Direct link: https://huggingface.co/Comfy-Org/MiniMax-H3/resolve/main/diffusion_models/minimax_h3_ref2va_pruned_fp8_scaled.safetensors?download=true

    -Other required files:

    nvfp4 Text Encoder: https://huggingface.co/Comfy-Org/MiniMax-H3/resolve/main/text_encoders/qwen3vl_32b_minimax_h3_nvfp4_awq.safetensors?download=true

    Video VAE: https://huggingface.co/Comfy-Org/MiniMax-H3/resolve/main/vae/minimax_h3_video_vae_fp16.safetensors?download=true

    Audio VAE: https://huggingface.co/Comfy-Org/MiniMax-H3/resolve/main/vae/minimax_h3_audio_vae_fp32.safetensors?download=true

    Description

    MiniMax H3 fl2va pruned fp8 scaled

    FAQ

    Comments (12)

    sekaiwlc07860Aug 4, 2026
    CivitAI

    What the difference between int8 and fp8,the latter more fast?

    Liant
    Author
    Aug 4, 2026· 3 reactions

    In simplified terms: FP8 is more precise, it uses floating-point. INT8 uses integral.

    WithoutOrdinaryAug 5, 2026· 3 reactions

    int8 can be slightly faster depending on hardware.

    thetoxiclambAug 8, 2026

    fp8 in 40 series and up , for others int8 then w4a8 then GGUF if both vram and ram is low

    yepeishengAug 5, 2026· 1 reaction
    CivitAI

    576*928 5秒视频在5060Ti 16G / 32G内存上跑了5:27

    DerpressionAug 5, 2026· 4 reactions
    CivitAI

    I generated one 3s clip with this, and instantly yeeted like 200 GB worth of LTX and Wan models. Holy diff O_O

    DaddyWolfgangAug 6, 2026

    Yeah. I can't even believe it myself... Like... Uncensored, works, very well, righ tout of the gate. I'm actually finding it unreal to be honest.

    DerpressionAug 6, 2026

    @DaddyWolfgang lol, same, and also the fact that you can do so much with basic ahh prompts @.@ Not having to feed it a novel just to control the camera feels like the years of my lifespan stolen by LTX are coming back

    DaddyWolfgangAug 6, 2026· 1 reaction
    CivitAI

    Guess I gotta eat crow. I never expected an uncensored model from that company. EVER. Glad I didn't put money on it being censored or not because I'd have lost.

    a0970384075547Aug 6, 2026

    LOL

    I_XXIVAug 7, 2026
    CivitAI

    On a 4070 12GB + 48GB - Generated a 5 seconds 544p clip in just 105 secs with the experimental Turbo LORA, upscaled to 720p (ish) with RTX Super Res. Not as fast as LTX, but the prompt adherance is WAYYYYY better.

    miradgamer971Aug 10, 2026

    @I_XXIV Hi! Please share your workflow.

    Checkpoint
    MiniMax H3
    by Liant
    Download (Beta) View on CivitAI
    checkpoint
    video
    base model
    video and audio

    Details

    Downloads
    1,161
    Platform
    CivitAI
    Platform Status
    Available
    Created
    8/4/2026
    Updated
    8/12/2026
    Deleted
    -

    Files

    minimax_h3_video_vae_fp16.safetensors

    Size:
    4.85 GB
    SHA256:
    7c1f131492e7eddacaac9069a61b81bdd39de5cc96561e677c5eab1cdce5e522

    Mirrors

    HuggingFace (9 mirrors)
    CivitAI (3 mirrors)

    minimaxH3PrunedFp8_fl2vaPrunedFp8Scaled_txt.safetensors

    Size:
    14.61 GB
    SHA256:
    35a88d51044231fe332301d7a62aa81e3f2cba62febeb446e2c1e3e0ef76f2c6

    Mirrors

    HuggingFace (3 mirrors)
    CivitAI (3 mirrors)

    minimax_h3_audio_vae_fp32.safetensors

    Size:
    577.22 MB
    SHA256:
    2550bd0ca3e91f0c46d66b359ed9f88df0aaa6e9438df43736427f5112ad2415

    Mirrors

    CivitAI (1 mirrors)

    minimaxH3PrunedFp8_fl2vaPrunedFp8Scaled.safetensors

    Size:
    19.52 GB
    SHA256:
    12944c1f7791637e7de12208aef04da82bd26b95271b1b47d817364315ade993

    Mirrors

    CivitAI (1 mirrors)