Text Models
Text Models are the AI language models available in the Text Node's generation panel. They take your prompt — along with any upstream text, images, or videos — and return a block of text.
Available models
| Model | Image input | Video input |
|---|---|---|
| Gemini 3-Flash | ✅ | ✅ |
| Gemini 3.1-Pro | ✅ | ✅ |
| GPT 5.4 | ✅ | ❌ |
The default model is Gemini 3.1-Pro.
Choosing a model
Gemini 3-Flash is the fastest option. Use it for straightforward text tasks — scene descriptions, character bios, simple rewrites — where speed matters more than nuanced output.
Gemini 3.1-Pro delivers higher quality output and handles complex instructions, long upstream contexts, and subtle creative direction more reliably. It also supports video input, making it the right choice when upstream Video Nodes are part of your workflow.
GPT 5.4 is an alternative high-quality option with a slightly different writing style and reasoning approach. Note that it does not accept video input — if any Video Node is connected upstream, those references will not be passed to GPT 5.4.
Note: If a generation fails due to too much upstream content, try reducing the number of connected upstream nodes or switching to a different model.
Updated about 2 months ago