ComfyUI Node
LLM Text Processor
A ComfyUI node in LLM Text Processor with 19 inputs and 3 outputs.
LLM Text Processor
- image
- RESPONSE
- REASONING
- PERF
◄model▾►
◄mmprojnone►
◄system_prompt▾►
◄promptDescribe this image in detail.►
◄max_tokens2048►
◄temperature0.70►
◄top_p0.80►
◄top_k20►
◄repeat_penalty1.00►
◄ctx_size8192►
◄memory_modeauto►
◄n_gpu_layers99►
◄n_cpu_moe_layers1►
◄seed1►
◄timeout_seconds300►
◄reasoningoff►
◄enable_processingtrue►
◄extra_args►
CategoryLLM Text Processor
Inputs (19)
| Name | Type | Default | Description |
|---|---|---|---|
| model | COMBO | GGUF model loaded from ComfyUI/models/LLM. mmproj files are hidden from this list. | |
| mmproj | COMBO | none | Vision projector GGUF. Required when one or more images are connected. |
| system_prompt | COMBO | System prompt preset from ComfyUI/models/LLM/prompts, or none. | |
| prompt | STRING | Describe this image in detail. | User prompt sent to the selected model. |
| max_tokens | INT | 20481–32768 | Maximum number of tokens to generate. |
| temperature | FLOAT | 0.700–2 | Sampling temperature. Lower is more deterministic. |
| top_p | FLOAT | 0.800–1 | Nucleus sampling threshold. |
| top_k | INT | 201–1000 | Top-K sampling cutoff. |
| repeat_penalty | FLOAT | 1.000–3 | Penalty applied to repeated tokens. |
| ctx_size | INT | 8192512–1048576 | Context window size in tokens. Use a value supported by the selected GGUF; larger context uses more VRAM. |
| memory_mode | COMBO | auto | Advanced memory placement mode: auto, gpu_layers, cpu_moe_layers, or gpu_and_cpu_moe_layers. |
| n_gpu_layers | INT | 99-1–999 | Used only in gpu_layers and gpu_and_cpu_moe_layers modes. Number of model layers to place on the GPU. |
| n_cpu_moe_layers | INT | 11–999 | Used only in cpu_moe_layers and gpu_and_cpu_moe_layers modes. Number of MoE layers to keep on the CPU. |
| seed | INT | 1-1–4294967295 | Random seed. Use -1 for a random seed. |
| timeout_seconds | INT | 30010–3600 | Maximum time to wait before generation is stopped. |
| reasoning | COMBO | off | Reasoning output mode. |
| imageopt | IMAGE | Optional image input. A single image or ComfyUI batch is passed to llama.cpp. | |
| enable_processingopt | BOOLEAN | true | When enabled, run normal node processing. When disabled, forward the input prompt directly as RESPONSE. |
| extra_argsopt | STRING | Optional advanced llama.cpp parameters. Leave empty for normal use. |
Outputs (3)
| Name | Type | Description |
|---|---|---|
| RESPONSE | STRING | Final model response with reasoning blocks removed. |
| REASONING | STRING | Extracted reasoning when present in model output. |
| PERF | STRING | llama.cpp prompt and generation speed. |