ComfyUI Node
[llama.cpp] Create Native Session
Keeps one Llama.cpp model resident until Llama.cpp Unload Session. Sessions left open during an interrupted workflow unload at prompt end. Place this before Start Loop and carry its session through End Loop.
[llama.cpp] Create Native Session
- model_profile
- hardware_profile
- reasoning
- speculative
- prefill_profile
- session
◄model_path[no GGUF models found]►
◄mmproj_path[none]►
◄custom_handlerauto►
◄n_ctx8192►
◄verbosefalse►
◄custom_chat_template—►
Categoryllama_cpp/session
Inputs (11)
| Name | Type | Default | Description |
|---|---|---|---|
| model_path | COMBO | [no GGUF models found] | 1 options: [no GGUF models found] |
| mmproj_path | COMBO | [none] | 1 options: [none] |
| custom_handler | COMBO | auto | A specific handler overrides Model Profile; auto keeps the profile's handler. |
| n_ctx | INT | 8192512–1048576 | — |
| verbose | BOOLEAN | false | — |
| model_profileopt | OLLAMA_IMAGE_LIST_LLAMA_CPP_MODEL_PROFILE | — | |
| custom_chat_templateopt | STRING | Optional custom Jinja template for this native session. When disconnected, the GGUF metadata template is used. | |
| hardware_profileopt | OLLAMA_IMAGE_LIST_LLAMA_CPP_HARDWARE_RUNTIME_PROFILE | — | |
| reasoningopt | OLLAMA_IMAGE_LIST_LLAMA_CPP_REASONING_CONFIG | — | |
| speculativeopt | OLLAMA_IMAGE_LIST_LLAMA_CPP_SPECULATIVE_CONFIG | — | |
| prefill_profileopt | OLLAMA_IMAGE_LIST_LLAMA_CPP_PREFILL_PROFILE | — |
Outputs (1)
| Name | Type | Description |
|---|---|---|
| session | OLLAMA_IMAGE_LIST_LLAMA_CPP_SESSION | — |