MiniMax-H3-Image in ComfyUI!
retained text
MiniMax has released **MiniMax-H3-Image**, and ComfyUI already has two native workflows for it. One checkpoint handles both generation from text and editing with image references. The release targets roughly **16 megapixels in a single pass**, with six sampling steps and no separate upscaler. [Official MiniMax model card](https://huggingface.co/MiniMaxAI/MiniMax-H3-Image) # Model downloads for ComfyUI These are the files selected by the official graphs. Paths are relative to ComfyUI/. * **Diffusion model, pruned INT8 ConvRot** · models/diffusion\_models/ · [Download checkpoint](https://huggingface.co/Comfy-Org/MiniMax-H3-Image/resolve/main/diffusion_models/minimax_h3_image_pruned_int8_convrot.safetensors) * **Qwen3-VL 32B text encoder, INT8 ConvRot** · models/text\_encoders/ · [Download encoder](https://huggingface.co/Comfy-Org/MiniMax-H3/resolve/main/text_encoders/qwen3vl_32b_minimax_h3_int8_convrot.safetensors) * **Image VAE, FP16** · models/vae/ · [Download VAE](https://huggingface.co/Comfy-Org/MiniMax-H3-Image/resolve/main/vae/minimax_h3_image_vae_fp16.safetensors) * **Qwen3.5 9B prompt expander, W4A8 GPTQ** · models/text\_encoders/ · [Download expander](https://huggingface.co/Comfy-Org/Qwen3.5/resolve/main/text_encoders/qwen3.5_9b_w4a8_gptq.safetensors) Update Your ComfyUI to 0.45 **IAMCCS**
document fetched 2026-11-11 00:42 · canonical https://reddit.com/r/comfyui/comments/1xssta9/minimaxh3image_in_comfyui