This repo provides a merge of lightx2v MiniMax prompt rewriter LoRA and it's base model Qwen3-VL-8B-Instruct and remapped to work with ComfyUI, includes an example workflow with system prompt.
See lightx2v's model card for details: https://huggingface.co/lightx2v/MiniMax-H3-Prompt-Rewriter-LoRA-8B
The benefit of this model is that it doesn't require custom nodes, it works with standard ComfyUI nodes out of the box, but it's a large model.
If you are willing to use custom nodes, you can also get a much smaller comfyui port of the LoRA here: https://huggingface.co/nynxz/Qwen3-VL-8B-ComfyUI
Inference Providers NEW
This model isn't deployed by any Inference Provider. 🙋 Ask for provider support