CivArchive
    AiAngelH3 - v2
    NSFW

    AiAngelH3 is a MiniMax H3 checkpoint merge for photorealistic video. Drop-in diffusion model for ComfyUI's Load Diffusion Model node.

    - v2 (recommended): AiAngelH3 v1 + Minimax-h3 Singularity v2 + H3 Eros Max beta7, transformer blocks merged 1:1:1. Better motion, good for everyday and action scenes.

    - v1: H3 Eros Max beta5 TURBO-hybrid 40% + RedCraft H3 A2A-RED beta1 60%.

    Settings, model files, RunPod and credits: see each version's notes.

    Credits: H3 Eros Max (tenstrip), RedCraft (AiMetatron), Minimax-h3 Singularity (AIGC_Singularity). Thank you all.

    License: H3 Eros Max allows only selling generated images and running on Civitai's own generator, so the same applies here: do not sell this model or merges of it, and do not run it on paid generation services outside Civitai. Running it yourself, locally or on your own rented GPU, is fine.

    NOTICE: This file has been modified from MiniMax H3 (a merge of MiniMax H3 model derivatives). MiniMax H3 is licensed under the MiniMax H3 Community License Agreement, Copyright © 2026 MiniMax. All Rights Reserved. By downloading or using this model you agree to that license and its Acceptable Use Policy: https://huggingface.co/MiniMaxAI/MiniMax-H3/blob/main/LICENSE

    The license does not cover the European Union, the United Kingdom, the Republic of Korea or the United States of America: do not download or use this model there. Anything you post publicly from it must be clearly labelled as AI-generated.

    Adults only. Do not use it to depict real people without their consent, or anyone under 18.

    Description

    v2 = AiAngelH3 v1 + Minimax-h3 Singularity v2 (AIGC_Singularity) + H3 Eros Max beta7 (tenstrip). Attention and MLP weights of every transformer block averaged 1:1:1 in int8; everything else is v1's.

    Recommended settings

    - euler / simple, 12 steps

    - LoRA minimax_h3_fl2v_turbo_8step_v1.0 at 1.0 (needed; without it v2 looks soft)

    - Fight scenes: add H3 Fight Flow LoRA at 0.65

    - 1080p: generate at 576x1024, then H3 latent upscaler (LBH-123) x1.875 -> 1088x1920, 4 steps res_multistep from 0.5, same prompt and seed

    - Needs the H3 text encoder (qwen3vl_32b_minimax_h3_nvfp4_awq), video VAE fp16 and audio VAE

    Speed (RTX PRO 6000, 10 s clip): ~80 s at 576x1024, +~155 s for 1080p.

    Samples use the same prompts and seeds as the v1 samples.

    RunPod: the free AI Angel ComfyPod template loads v2 by default (MODELS=h3core,aiangelh3,h3upscaler). For v1 add aiangelh3v1.

    Template: https://github.com/ThepExcel/AiAngelComfyPod

    New to RunPod: https://runpod.io?ref=fbrjus7u

    FAQ

    Checkpoint
    MiniMax H3

    Details

    Downloads
    310
    Platform
    CivitAI
    Platform Status
    Available
    Created
    11/18/2026
    Updated
    11/20/2026
    Deleted
    -

    Files

    AiAngelH3-v2-int8.safetensors

    Mirrors