Comfy-Org republishes MiniMax-H3-Image weights for ComfyUI
Original title: Comfy-Org/MiniMax-H3-Image — new model trending #30 on Hugging Face
The pack adds int8 and NVFP4 files plus Qwen encoders, under a license that restricts high-revenue hosted use.
Comfy-Org published repackaged MiniMax H3 Image files for ComfyUI, pointing to the original MiniMaxAI repository. The layout covers a pruned int8 diffusion model, Qwen3.5 and Qwen3-VL text encoders, and an fp16 VAE. A separate MiniMax H3 Image license requires prior written authorization if annualized revenue exceeds 1 million USD or if the user offers hosted services to third parties.
- Prompt enhancement is recommended with qwen3.5_9b_w4a8_gptq in the Generate Text node.
- Text encoders include qwen3vl_32b builds in int8 and NVFP4 AWQ.
- Hosted use above 1 million USD annualized revenue needs separate MiniMax authorization.
- Products
- ComfyUIMiniMax H3 Image
- AI models
- MiniMax-H3-Imageqwen3.5_9b_w4a8_gptq
From the original article
1 min read at huggingface.coOriginal model repository: https://huggingface. co/MiniMaxAI/MiniMax-H3-Image For prompt enhancement, we recommend to use the model `qwen3. 5_9b_w4a8_gptq` with the Generate Text node.
By Comfy-Org · huggingface.co
A short excerpt; the article and its images belong to the publisher. Summary, key facts and assessment above are AI-written from the article and can be wrong.