Comfy-Org/MiniMax-H3

1---2license: other3license_name: minimax-h3-community-license-agreement4license_link: https://huggingface.co/MiniMaxAI/MiniMax-H3/blob/main/LICENSE5tags:6- comfyui7- diffusion-single-file8base_model:9- MiniMaxAI/MiniMax-H310---11 12# MiniMax H313 14Repackaged model files for ComfyUI.15 16Original model repository:17 18- https://huggingface.co/MiniMaxAI/MiniMax-H319- https://huggingface.co/lightx2v/Minimax-h3-Turbo20- https://huggingface.co/alibaba-pai/MiniMax-H3-Fun-Controlnet-Union21- https://huggingface.co/Kijai/MiniMax-H3-experimental22 23The Qwen3-VL-32B `nvfp4_awq` quant is converted from: https://huggingface.co/cybermotaz/Qwen3-VL-32B-Instruct-NVFP424 25This `nvfp4` text encoder does not require Blackwell GPU to use.26 27For diffusion models prefer `int8_convrot` if you are able to use pytorch with cu130.28 29`fp8_scaled` should only be used if you cannot use `int8_convrot`.30 31---32 33Place the files in the following folders:34 35```36📂 ComfyUI/37├── 📂 models/38│   ├── 📂 diffusion_models/39│   │   ├── minimax_h3_fl2va_bf16.safetensors40│   │   ├── minimax_h3_fl2va_int8_convrot.safetensors41│   │   ├── minimax_h3_fl2va_pruned_bf16.safetensors42│   │   ├── minimax_h3_fl2va_pruned_int8_convrot.safetensors43│   │   ├── minimax_h3_fl2va_pruned_fp8_scaled.safetensors44│   │   ├── minimax_h3_ref2va_bf16.safetensors45│   │   ├── minimax_h3_ref2va_int8_convrot.safetensors46│   │   ├── minimax_h3_ref2va_pruned_bf16.safetensors47│   │   ├── minimax_h3_ref2va_pruned_int8_convrot.safetensors48│   │   └── minimax_h3_ref2va_pruned_fp8_scaled.safetensors49│   ├── 📂 text_encoders/50│   │   ├── qwen3vl_32b_minimax_h3_bf16.safetensors51│   │   ├── qwen3vl_32b_minimax_h3_int8_convrot.safetensors52│   │   └── qwen3vl_32b_minimax_h3_nvfp4_awq.safetensors53│   ├── 📂 loras/54│   │   ├── minimax_h3_fl2v_turbo_4step_v1.0_768p_comfyui_bf16.safetensors55│   │   ├── minimax_h3_fl2v_turbo_8step_v1.0_comfyui_bf16.safetensors56│   │   └── minimax_h3_ref2v_turbo_4step_v0.1_comfyui_bf16.safetensors57│   ├── 📂 vae/58│   │   ├── minimax_h3_audio_vae_fp32.safetensors59│   │   ├── minimax_h3_audio_vae_fp32.safetensors60│   │   └── minimax_h3_video_vae_int8_convrot.safetensors61│   ├── 📂 model_patches/62│   │   ├── minimax_h3_fun_controlnet_union_pruned_bf16.safetensors63│   │   └── minimax_h3_fun_controlnet_union_pruned_int8_convrot.safetensors64│   ├── 📂 embeddings/65│   │   ├── minimaxh3_art_is_explosion.safetensors66│   │   ├── minimaxh3_blooming_flowers.safetensors67│   │   ├── minimaxh3_bullet_time.safetensors68│   │   ├── minimaxh3_dark_magic.safetensors69│   │   ├── minimaxh3_fire_breath.safetensors70│   │   ├── minimaxh3_four_seasons.safetensors71│   │   ├── minimaxh3_kiss_camera.safetensors72│   │   ├── minimaxh3_spiral_ascent.safetensors73│   │   ├── minimaxh3_storm_magic.safetensors74│   │   └── minimaxh3_truman_show.safetensors75```76 77## Workflows78 79- I2V：https://github.com/Comfy-Org/workflow_templates/blob/main/templates/video_minimax_h3_i2v.json80- T2V：https://github.com/Comfy-Org/workflow_templates/blob/main/templates/video_minimax_h3_t2v.json81- R2V：https://github.com/Comfy-Org/workflow_templates/blob/main/templates/video_minimax_h3_r2v.json82 83Doc: https://docs.comfy.org/tutorials/video/minimax/minimax-h384 85## Embeddings86 87To use the embeddings in a text prompt, invoke them in the CLIPTextEncode node by88combining `embedding:` with the filename of the embedding like this for `minimaxh3_art_is_explosion`:89```90embedding:minimaxh3_art_is_explosion91```92Just exchange the `minimaxh3_art_is_explosion` part with the filename of the embedding you want to use.