⚠️ [Deprecated] Review Images

DeepGen_I2T is the old name for the pack's vision-language node - the one that reads up to ten of your images and writes back text. When you load a legacy workflow and see "⚠️ [Deprecated] Review Images", this is it, and it's not about to disappear. It's the same class that now ships as DeepGen_I2T0, registered under its pre-rename key with the deprecated category and flag for visibility. The rename happened when the pack standardized its node names around version 1.2.1: current nodes got 0/X suffixes, and old keys stayed as aliases so existing graphs keep loading.

What it does

Wire in up to ten images (image_1–image_10), give a prompt that says what you want done - compare the batch, caption a render, critique against your reference - and a cloud VLM answers. Model options are Gemini 3 Pro, Claude Opus 4.6 / Sonnet 4.6, GPT-5.4, Gemini 3 Flash, and Gemini 3.1 Flash Lite. Outputs are output (STRING), model (STRING), and total_credits_used (FLOAT). Full walkthrough in the DeepGen_I2T0 article.

Install

Same single pack, same steps:

cd ComfyUI/custom_nodes
git clone https://github.com/deepiksdev/ComfyUI-DeepGen-API.git
cd ComfyUI-DeepGen-API
pip install -r requirements.txt

Restart ComfyUI and add your DeepGen API key (deepgen.app) in Settings → "DeepGen API Key", saved to ComfyUI/user/deepgen/config.json. No heavy dependencies - requests and opencv-python only.

What to do with it

New workflows: use DeepGen_I2T0. Old workflows referencing DeepGen_I2T keep working untouched - that's the whole point of the compatibility alias. If you're refactoring a legacy graph, swap the node for DeepGen_I2T0 and reconnect; the inputs and outputs are identical, so it's a fast swap. And if a stable graph is fine as it is, leave it. Deprecated here means renamed with backward compatibility, not condemned - the warning flag is the pack telling you the name is old, not the node.