LayerUtility: ZhipuGLM4V(Advance)

ZhipuGLM4V sends an image to Zhipu AI's vision API and gets a text description back. It's the vision counterpart to the pack's ZhipuGLM4 text node, and the appeal is the same reason it's worth knowing about at all: unlike the Gemini or DeepSeek nodes elsewhere in this pack, Zhipu's entry-level vision model is free to use, no trial period, no credit card.

A quick reality check on who you're actually calling: Zhipu AI is the Tsinghua University spin-out (originally Zhipu, internationally rebranded as Z.ai in 2025) that also built CogVideoX and the GLM model line, and it IPO'd in Hong Kong in January 2026 as China's first publicly listed frontier AI lab. Worth knowing because this particular node is hitting their hosted API, not open weights - Z.ai ships plenty of open-source models elsewhere, but GLM-4V here is a paid service with a free tier, same as Gemini is for Google.

The inputs and outputs that matter

• image - required, what you want described.
• model - five options: glm-4.5v, glm-4.1v-thinking-flashx, glm-4v-flash, glm-4v, glm-4v-plus. Of these, glm-4v-flash is the one confirmed free in the Zhipu documentation the README points to; the others (including the newer glm-4.5v and the "thinking" variant, both added to the model list since the README was last updated) are paid, generally in exchange for better reasoning and image understanding.
• user_prompt (default "describe this image", multiline) - this is your actual instruction. Change it to ask for something more specific than a generic description - a caption in a particular style, a list of objects, whatever you need.

Output is a single text string.

Installing it

ComfyUI Manager: search "ComfyUI Layer Style Advance". Manual:

cd ComfyUI/custom_nodes
git clone https://github.com/chflame163/ComfyUI_LayerStyle_Advance.git

Run install_requirements.bat (portable) or install requirements.txt yourself, then restart.

API key setup, the part people actually get stuck on: register for a free key at bigmodel.cn/usercenter/proj-mgmt/apikeys, then rename api_key.ini.example (in the plugin's own folder) to api_key.ini, open it in a text editor, and fill in zhipu_api_key= with your key. This file is shared across every Zhipu node in the pack - do it once and ZhipuGLM4 and ZhipuGLM4V both pick it up.

Common issues

Node errors or silently fails with no clear message. Check api_key.ini exists (not just the .example version) and actually has your key in the right field. This is the single most common way any of this pack's API nodes "don't work."

Picked glm-4v or glm-4v-plus and got billed unexpectedly. Only glm-4v-flash is free per Zhipu's own docs - the rest are metered. If you're testing this node for the first time, start on flash.

Requests time out or fail from certain networks. Zhipu's API is hosted in China; depending on where you're running ComfyUI, latency or connectivity can be worse than hitting Gemini or OpenAI's endpoints. If you're getting consistent connection errors rather than API errors, that's the first thing to suspect.

Output reads generic no matter what you ask. The default user_prompt is deliberately bland ("describe this image"). Be specific about what you want back - GLM-4V responds to instruction the same way any vision-chat model does, it just needs you to actually ask.