ComfyUI Node
Unsloth Generate
Text (and vision) generation via an Unsloth Desktop OpenAI-compatible backend. Connect an Unsloth Connectivity node to set the server, key, and model.
Unsloth Generate
- connectivity
- options
- images
- meta
- result
- thinking
- meta
◄systemYou are an AI artist.►
◄promptWhat is art?►
◄thinkfalse►
◄format▾►
CategoryUnsloth
Inputs (8)
| Name | Type | Default | Description |
|---|---|---|---|
| system | STRING | You are an AI artist. | System prompt — sets the role and general behavior of the model. |
| prompt | STRING | What is art? | User prompt. For vision tasks you can refer to the input image as 'this image'. |
| think | BOOLEAN | false | If enabled, the model does a thinking pass first. The thinking is returned as a separate output. Some models do not support this. |
| format | COMBO | 'json' requests a JSON object response (response_format). Show the model example outputs in the system prompt for best results. | |
| connectivityopt | UNSLOTH_CONNECTIVITY | Connection to set the server, key, model, and keep-alive. Required unless provided via 'meta'. | |
| optionsopt | UNSLOTH_OPTIONS | Advanced sampling options. | |
| imagesopt | IMAGE | Optional image or batch for vision tasks. The model must support vision. | |
| metaopt | UNSLOTH_META | Chain multiple Generate nodes; connectivity and options are passed along (and can be overridden). |
Outputs (3)
| Name | Type | Description |
|---|---|---|
| result | STRING | — |
| thinking | STRING | — |
| meta | UNSLOTH_META | — |