CivArchive
    Anima Zenkai - Preview3
    NSFW
    Preview 128423885
    Preview 128423570
    Preview 128423572
    Preview 128423586
    Preview 128423605
    Preview 128423612
    Preview 128423614
    Preview 128423619
    Preview 128424002
    Preview 128424008
    Preview 128424010
    Preview 128424014
    Preview 128424015
    Preview 128424096
    Preview 128424357

    Based on Anima Preview 2

    Anima is a 2-billion parameter text-to-image model developed by CircleStone Labs in collaboration with Comfy Org. Built on the NVIDIA Cosmos architecture and utilizing the Qwen text encoder, it serves as a specialized foundation for high-end anime and artistic generation.

    Model recipe

    Merge of AnimaYume_v0.3 + AnimaCatTower_v0.4 + own trained LoRAs

    Merging the models:

    Merging the loras:

    Workflows are embedded in the images.

    • Steps: 25

    • CFG Scale: 5.0

    • Sampler: euler_ancestral

    • Scheduler: simple

    ComfyUI Workflow

    Description

    FAQ

    Checkpoint
    Anima
    by litch264187
    Download (Beta) View on CivitAI
    anime
    base model

    Details

    Downloads
    177
    Platform
    CivitAI
    Platform Status
    Available
    Created
    4/24/2026
    Updated
    10/2/2026
    Deleted
    -

    Files

    animaZenkai_preview3_txt.safetensors

    Size:
    1.11 GB
    SHA256:
    1abdf1539dfce592439f53d18e4e29528bfa9e2a38b0e2a44a8b4095f118ff84

    Mirrors

    CivitAI (1 mirrors)

    animaZenkai_preview3.safetensors

    Size:
    3.90 GB
    SHA256:
    d32ab95c45f036ad487e97ab99ba52e1e93f233682b4d8563834e6da7db35b0b

    Mirrors

    CivitAI (1 mirrors)
    TensorArt (1 mirrors)
    TensorHub (1 mirrors)

    qwen_image_vae.safetensors

    Size:
    242.05 MB
    SHA256:
    a70580f0213e67967ee9c95f05bb400e8fb08307e017a924bf3441223e023d1f

    Mirrors

    HuggingFace (314 mirrors)
    CivitAI (257 mirrors)
    TensorFiles (1 mirrors)
    ModelScope (3 mirrors)
    ModelScope CN (6 mirrors)