ComfyUI Node
Local chat completions
A ComfyUI node in ✨ LF Nodes/LLM with 10 inputs and 2 outputs.
Local chat completions
- image
- ui_widget
- text
- response_json
◄prompt►
◄urlhttp://127.0.0.1:1234/v1/chat/completions►
◄system_message►
◄model►
◄temperature0.2►
◄max_tokens4096►
◄timeout120►
◄reasoningauto►
Category✨ LF Nodes/LLM
Inputs (10)
| Name | Type | Default | Description |
|---|---|---|---|
| prompt | STRING | User prompt sent to the local chat-completions endpoint. | |
| url | STRING | http://127.0.0.1:1234/v1/chat/completions | Local OpenAI-compatible /v1/chat/completions or LM Studio native /api/v1/chat endpoint. |
| system_messageopt | STRING | Optional system instruction sent before the user prompt. | |
| imageopt | IMAGE | Optional ordered reference images for a vision-capable model. | |
| modelopt | STRING | Optional. Blank LM Studio native chat requests use the sole loaded LLM instance. | |
| temperatureopt | FLOAT | 0.20–2 | Controls response randomness. |
| max_tokensopt | INT | 4096 | Response token budget, including any model reasoning. |
| timeoutopt | INT | 120 | Request timeout in seconds. |
| ui_widgetopt | LF_CODE | — | |
| reasoningopt | COMBO | auto | LM Studio native reasoning control. auto uses the model default; off/on require /api/v1/chat. |
Outputs (2)
| Name | Type | Description |
|---|---|---|
| text | STRING | Exact answer text returned by the local model. |
| response_json | JSON | Complete JSON response returned by the endpoint. |