😼> Select LMS Model

A tiny convenience node that ships alongside the main LMStudio node in the same pack. Its entire job: read your model list from a plain text file, show you a dropdown, and hand back the one you picked as a STRING you can wire straight into the model_identifier input of the > LMStudio node.

Why you'd bother

The main node's model_identifier is a raw text field, and LM Studio's identifiers are long, fully-qualified names like Qwen3.6-35B-A3B-Instruct-... that are begging to be mistyped. If you switch models often - a vision model for image input here, a prompt-expander there - a dropdown turns "request failed, model not found" into a two-click change. It also keeps the model name in one place instead of buried in a workflow node, which matters if you're sharing workflows with someone who has no idea what to type in that field.

How it works

On startup the node reads every non-empty line of models.txt in the pack's lms_config folder. Each line becomes one entry in the model_id dropdown. Pick one and it comes out the other side unchanged. That's the whole mechanism - it's a lookup table with a UI, not a registry query.

One catch: the file is read when ComfyUI loads the node class, so after you edit it you need to restart ComfyUI (or refresh the node list) before new entries appear. And out of the box the file contains a single line - "Please see tooltip." - so your first dropdown will have one useless entry. That's not a bug, that's the author leaving a sticky note. Time to edit the file.

Setting it up

# ComfyUI/custom_nodes/YANC_LMStudio/lms_config/models.txt
Qwen3.6-35B-A3B-Instruct
Qwen3.6-35B-VL-Instruct
llama-3.3-70b-instruct

One model per line, no commas, no extra text. The names must match what LM Studio shows in its own model dropdown exactly - case and hyphens included - because that's the string the main node sends to the API, and it won't fuzzy-match for you.

Install and gotchas

Same story as its sibling node: clone the repo or install via ComfyUI Manager, then pip install -r requirements.txt (just lmstudio and msgspec), restart ComfyUI. Both nodes land under the YANC → LMStudio submenu. Nothing to download - no models, no config beyond models.txt.

Here's my honest take: this node is optional. If you only ever use one model, type its name into the main node and skip this entirely. Reach for it when you're juggling several models or handing workflows to people who'll be lost staring at an empty identifier field. Just remember the list is hand-maintained - add or remove models in LM Studio and this dropdown won't update itself, because it never looks at LM Studio. Plain-text file, manual sync. That's the tradeoff for being able to edit it with any editor.

Troubleshooting

Dropdown only shows "Please see tooltip." - you haven't added your models yet; that's the placeholder line.
New models not appearing - restart ComfyUI after saving; the file is only read once at node load.
Request fails though the name looks right - the identifier must match LM Studio exactly, including case and punctuation. Copy it from LM Studio's dropdown rather than retyping it.