ComfyUI Node
Ollama LLM Loader (TJ)
Ollama-only local LLM caller with TJ_NODE embedded Set/Get support.
Ollama LLM Loader (TJ)
- image
- response
- thought_process
- raw_response
◄get_name▾►
◄setnode_name►
◄server_urlhttp://127.0.0.1:11434►
◄model►
◄system_prompt►
◄user_prompt►
◄thinkingfalse►
◄seed1►
◄seed_modefixed►
◄temperature0.7►
◄top_p0.9►
◄repeat_penalty1.1►
◄model_memoryUnload after run►
◄keep_minutes5►
◄comfy_vram_policyAuto: unload only before first LLM call►
◄prompt_in—►
Category ✨ TJ_Node/LLM
Inputs (17)
| Name | Type | Default | Description |
|---|---|---|---|
| get_name | COMBO | 1 options: (none) | |
| setnode_name | STRING | — | |
| server_url | STRING | http://127.0.0.1:11434 | — |
| model | COMBO | 1 options: | |
| system_prompt | STRING | — | |
| user_prompt | STRING | — | |
| thinking | BOOLEAN | false | — |
| seed | INT | 10–4294967295 | — |
| seed_mode | COMBO | fixed | 4 options: fixed, increment, decrement, randomize |
| temperature | STRING | 0.7 | — |
| top_p | STRING | 0.9 | — |
| repeat_penalty | STRING | 1.1 | — |
| model_memory | COMBO | Unload after run | 3 options: Unload after run, Keep for minutes, Keep loaded |
| keep_minutes | STRING | 5 | — |
| comfy_vram_policy | COMBO | Auto: unload only before first LLM call | 3 options: Auto: unload only before first LLM call, Always unload before each LLM call, Never unload before LLM call |
| imageopt | IMAGE | — | |
| prompt_inopt | STRING | — |
Outputs (3)
| Name | Type | Description |
|---|---|---|
| response | STRING | — |
| thought_process | STRING | — |
| raw_response | STRING | — |