Google Gemini
Still works, but it's the deprecated one now
- images
- audio
- video
- files
- STRING
This is the first-generation Gemini text node, and it's got a big yellow flag: the source marks it deprecated in favor of the newer Google Gemini (V2). It still runs, it still works, and if you've got an old workflow built on it you don't need to panic. But if you're starting fresh, reach for V2 - it's the one getting the new models and the per-model controls. Consider this article the history lesson and the migration guide.
The mechanism
Same family mechanics: prompt plus optional multimodal context goes to Gemini's Generate Content API through Comfy's proxy, billed through your Comfy account in credits, price badge on the node. What made this node distinctive when it landed in May 2025 is the breadth of context types - it takes images, audio, video, and files (from Gemini Input Files) all in one request. That's a genuinely multimodal node: describe a clip, caption a frame, summarize an audio file.
The inputs that matter
prompt- the text, multiline.model- a fixed combo (gemini-3-1-pro by default, plus gemini-3-1-flash-lite, gemini-2.5-pro/flash, and the 3-pro-preview labels). One caveat worth knowing: some of those dropdown names are remapped under the hood - "gemini-3-pro-preview" quietly becomes the newer "gemini-3.1-pro-preview" because Google keeps deprecating the labels. The list can lag Google's renames; if a name feels stale, that's why.seed- default 42, and the tooltip is unusually honest: a fixed seed makes a best effort at the same response, determinism isn't guaranteed, and changing temperature or model breaks it anyway.images/audio/video/files- the optional multimodal context, each a single slot (use Batch Images on the image side for multiples).system_prompt- optional foundational instructions.
Output is a STRING.
Getting it and the honest verdict
Ships with ComfyUI core; the Gemini nodes have been in since May 2025. Same account-and-credits catch as the whole family, same non-determinism, needs internet.
Where the deprecation bites: V2 exists precisely because Google's newer models expose knobs this node can't express - thinking levels, temperature, a much bigger token budget. This node's model list is frozen in time while V2's dynamic combo tracks the current lineup. It's not broken, and it's not going to vanish from existing saves, but every new model, every new control, and every fix lands on V2. If you're still on this one, the migration is a two-minute cut-and-paste of the same inputs onto the V2 node. Do it once and stop reading deprecation notices.
Inputs (8)
| Name | Type | Default | Description |
|---|---|---|---|
| prompt | STRING | Text inputs to the model, used to generate a response. You can include detailed instructions, questions, or context for the model. | |
| model | COMBO | gemini-3-1-pro | The Gemini model to use for generating responses. |
| seed | INT | 420–18446744073709550000 | When seed is fixed to a specific value, the model makes a best effort to provide the same response for repeated requests. Deterministic output isn't guaranteed. Also, changing the model or parameter settings, such as the temperature, can cause variations in the response even when you use the same seed value. By default, a random seed value is used. |
| imagesopt | IMAGE | Optional image(s) to use as context for the model. To include multiple images, you can use the Batch Images node. | |
| audioopt | AUDIO | Optional audio to use as context for the model. | |
| videoopt | VIDEO | Optional video to use as context for the model. | |
| filesopt | GEMINI_INPUT_FILES | Optional file(s) to use as context for the model. Accepts inputs from the Gemini Generate Content Input Files node. | |
| system_promptopt | STRING | Foundational instructions that dictate an AI's behavior. |
Outputs (1)
| Name | Type | Description |
|---|---|---|
| STRING | STRING | — |