AI_Chooserđ§¸
The little node that tells Style Prompt which model to use
- AI_Selection
AI Chooser is a helper node with exactly one job: pick an AI service and a model, and pass that choice to the Style Prompt (Enhancer) node. That's it. It doesn't call anything, doesn't generate anything, and doesn't cost anything on its own. If you found this page trying to figure out why your Style Prompt workflow won't run, this is almost certainly the missing piece.
Why it exists
Originally the Style Prompt node only talked to ChatGPT, and you picked the GPT model right on the node. Then the author added support for Groq, Anthropic, and Gemini. Rather than cram four service pickers and four model dropdowns onto Style Prompt itself, he split that selection out into its own small node. So now Style Prompt has a single AI_Selection input, and AI Chooser is what fills it. Clean separation, but it trips up anyone who drags a Style Prompt onto the canvas and expects it to work solo.
How it works
You set AI_Service to one of ChatGPT, Groq, Gemini, or Anthropic, then pick the matching model from that service's dropdown. The node bundles your choice into an AI_Selection dictionary and outputs it. You wire that single output into the AI_Selection input on Style Prompt. Done.
The inputs
- AI_Service: the service you want. Four options: ChatGPT, Groq, Gemini, Anthropic.
- ChatGPT_model / Groq_model / Google_Gemini_model / Anthropic_model: the model dropdown for each service. These populate from your API account once your key is set, which is why they read as empty ("none") until Plush can see a valid key. Only the dropdown matching your chosen service matters.
The single output is AI_Selection, a dictionary. It goes into Style Prompt and nowhere else.
Installing it
AI Chooser ships inside Plush-for-ComfyUI, so you don't install it separately. Get the pack through ComfyUI Manager (search "Plush", install Plush-for-ComfyUI, restart), or manually with cd ComfyUI/custom_nodes && git clone https://github.com/glibsonoran/Plush-for-ComfyUI.git, then pip install -r requirements.txt in the folder and restart.
Common issues
If the model dropdowns are empty, your API key isn't being seen. Plush reads keys from environment variables (OAI_KEY or OPENAI_API_KEY for ChatGPT, GROQ_API_KEY for Groq, ANTHROPIC_API_KEY for Anthropic, GEMINI_API_KEY for Gemini), and setting one usually needs a machine reboot before it takes. No key means no model list.
One version-drift note straight from the pack's changelog: when the author reworked the AI Chooser and Advanced Prompt Enhancer UIs, existing workflows that already had these nodes could throw errors after an update. The fix is to right-click the node and hit "Reload Node" (or just re-select your service and model). If an old workflow suddenly complains about this node after you updated Plush, that's what's happening. Groq is worth a mention as the low-friction way to test the whole chain, since its key is free.
Inputs (5)
| Name | Type | Default | Description |
|---|---|---|---|
| AI_Service | COMBO | ChatGPT | 4 options: ChatGPT, Groq, Gemini, Anthropic |
| ChatGPT_model | COMBO | 1 options: none | |
| Groq_model | COMBO | 1 options: none | |
| Google_Gemini_model | COMBO | 1 options: none | |
| Anthropic_model | COMBO | 1 options: none |
Outputs (1)
| Name | Type | Description |
|---|---|---|
| AI_Selection | DICTIONARY | â |