Nodes/ComfyUI-988/LM Studio 988
ComfyUI Node

LM Studio 988

LM Studio 988 — a node for integrating with LM Studio's local LLM/VLM capabilities.

By kajan988·Created 2 months ago·Updated 21 days ago· 0
LM Studio 988
  • image1
  • image2
  • image3
  • image4
  • response
  • reasoning
  • troubleshooting
system_messageYou are a helpful assistant.
default_system_message🚫OFF
prompt
model_selection-- Custom (enter below) --
custom_model_name
max_tokens1024
temperature0.70
seed0
image_resizeMedium (768px)
draft_model_selection-- Custom (enter below) --
custom_draft_model
top_p1.00
top_k0
repeat_penalty1.00
reasoning_modeAuto-detect (recommended)
custom_open_tag<think>
custom_close_tag</think>
unload_all_modelsfalse
empty_cuda_cachefalse
refresh_modelsfalse
Category🦇988/LM Studio

Inputs (24)

NameTypeDefaultDescription
system_messageSTRINGYou are a helpful assistant.System prompt that defines the LLM's role and behavior.
default_system_messageCOMBO🚫OFFSelect a predefined system message template to append after your custom system message.
promptSTRINGThe user prompt to send to the LLM.
model_selectionCOMBO-- Custom (enter below) --Select a model from LM Studio. Select 'Custom' to manually enter a model identifier.
custom_model_nameSTRINGManual model identifier. Only used when 'Custom' is selected above.
max_tokensINT10241–131072Maximum OUTPUT tokens for the response.
temperatureFLOAT0.700–2Controls randomness. Lower = focused, Higher = creative.
seedINT00–18446744073709550000Seed for reproducibility.
image_resizeoptCOMBOMedium (768px)Resize images before processing. Smaller = faster inference.
image1optIMAGEFirst image input for vision models (VLMs).
image2optIMAGESecond image input for multi-image VLMs.
image3optIMAGEThird image input for multi-image VLMs.
image4optIMAGEFourth image input for multi-image VLMs.
draft_model_selectionoptCOMBO-- Custom (enter below) --Optional draft model for speculative decoding.
custom_draft_modeloptSTRINGManual draft model identifier. Leave empty to disable.
top_poptFLOAT1.000–1Nucleus sampling threshold. 1.0 = disabled.
top_koptINT00–500Top-K sampling. 0 = disabled. Recommended: 20-40 for thinking models.
repeat_penaltyoptFLOAT1.000–2Penalizes repeated tokens. 1.0 = disabled.
reasoning_modeoptCOMBOAuto-detect (recommended)How to extract reasoning/thinking from model output.
custom_open_tagoptSTRING<think>Custom opening tag for reasoning extraction.
custom_close_tagoptSTRING</think>Custom closing tag for reasoning extraction.
unload_all_modelsoptBOOLEANfalseUnloads ComfyUI models (SD, VAE, etc.) from VRAM.
empty_cuda_cacheoptBOOLEANfalseClears the CUDA cache.
refresh_modelsoptBOOLEANfalseToggle ON to re-fetch the model list from LM Studio.

Outputs (3)

NameTypeDescription
responseSTRING
reasoningSTRING
troubleshootingSTRING