🤖 Gemini Model Selector

This node does exactly one thing: it's a dropdown of Gemini model names that outputs the selected name as a string. It makes no API call, needs no key, and can't look at an image. What it's for is being the single control point for the pack's Gemini API nodes - connect its output to the model_override input of the ArchAi3D_Gemini node and you can switch between Gemini models without digging into that node's settings.

The model list (and what it tells you)

As of early 2026 the dropdown carries the current lineup:

• Gemini 3 previews - gemini-3-pro-preview, gemini-3-pro-image-preview, gemini-3-flash-preview. The image-preview variant is the interesting one if you're using Gemini for image work.
• Gemini 2.5 (stable GA) - gemini-2.5-flash, gemini-2.5-flash-lite, gemini-2.5-flash-image, gemini-2.5-pro.
• Gemini 2.0 - gemini-2.0-flash, gemini-2.0-flash-lite, flagged in the source as retiring March 2026.

The name input (default gemini_model) is the same web-interface identifier the pack's other input nodes use - it lets web frontends generate a form dropdown from your workflow. Output is a single model string.

Why this pattern exists

Same rationale as ArchAi3D_Conditioning_Balance: coordination. The pack's Gemini node (ArchAi3D_Gemini) is a full VLM call node that accepts a model_override string, and if you've got several Gemini nodes in a workflow, you want one dropdown that controls all of them. It also keeps model names out of the "hardcoded widget" trap - update the list once here and every downstream node follows.

The honest caveats

This node is just the selector; the actual model call happens in ArchAi3D_Gemini, which needs google-genai installed and a working API key in your environment. If you're not using the pack's Gemini integration at all, this node is dead weight - it outputs a string nobody reads. And a word on the bigger picture: Gemini is a closed, paid API (the KB's closed-source-models essay has the full context), so any workflow with it in has a dependency on Google's service and billing. For local Qwen-Image editing you don't need it at all.

Install

Part of the ArchAi3D Qwen pack:

cd ComfyUI/custom_nodes
git clone https://github.com/amir84ferdos/ComfyUI-ArchAi3d-Qwen.git

The node itself needs nothing (stdlib only) - google-genai is required only if you run the actual Gemini API node. Restart ComfyUI, find it under ArchAi3d/Edit/VLM. Pack license: free personal, paid commercial.