We’re excited to announce the public release of ComfyUI-MiniCPM, a powerful custom node for ComfyUI that integrates MiniCPM-V vision-language models (both Transformers and GGUF) into your local workflows.

What it can doComfyUI-MiniCPM enables high-quality visual instruction-following for tasks such as:

Image and video captioning

Object and scene identification

Visual analysis and explanation

Key Features

Support for both Transformers and GGUF models

Automatic model downloading (CPU & GPU variants)

Prompt types: Describe, Analyze, Identify, Explain, Caption, etc.

Speed/memory optimization options for efficient local usage

https://github.com/1038lab/ComfyUI-MiniCPM

If this node is helpful to you or if you like Our work, please give a ⭐ on Github, It will be the greatest encouragement for our efforts!