Nodes/SimpleSyrup/Grounded SAM Model Info
ComfyUI Node

Grounded SAM Model Info

Ask the Pack Which Detector Models You're Actually Using

By Artificial-Sweetener·Created 3 months ago·Updated 12 days ago· 2
Grounded SAM Model Info
    • model_info
    sam_modelsam_hq_vit_b (379MB)
    grounding_dino_modelGroundingDINO_SwinT_OGC (694MB)

    Every now and then you inherit a workflow and realize you don't know what models it's actually running under the hood. Grounded SAM Model Info is the node that answers that question before the question becomes a bug: pick a SAM model and a GroundingDINO model from the dropdowns, and it returns a JSON string describing both - model sources, what they are, the metadata the pack knows about them. It's a diagnostic/utility node, not a loader; it doesn't load anything, it just tells you about the choices.

    How it works

    Two combo inputs, one text output:

    • sam_model (default sam_hq_vit_b, 379MB): the SAM variant. The catalog covers the SAM-HQ line (vit_b/l/h) plus MobileSAM, so the dropdown reflects what the pack can actually fetch and run.
    • grounding_dino_model (default GroundingDINO_SwinT_OGC, 694MB): the detector side.

    The model_info output is a STRING of JSON describing the selected SAM and GroundingDINO sources - think of it as the pack's internal catalog entry for each choice, exposed so you can read it. Wire it into a text display or a note node, or just use it to sanity-check that a shared workflow's expected models match what you have installed.

    Why you'd bother

    Two honest uses. First, workflow forensics: someone shares a graph, you open it, and the SAM loader defaults to something you've never heard of - run this node with the same selections and you get the authoritative description without digging through the pack source. Second, sizing: the dropdown labels carry sizes (sam_hq_vit_h is 2.57GB, sam_hq_vit_b is 379MB), and the JSON tells you where each one downloads from, which is exactly the info you need when deciding whether to switch the heavyweight variant for the light one on a smaller GPU.

    Beyond that, it's a passthrough utility with no state and no side effects. If you never need it, you'll never miss it. If you're the kind of person who likes knowing what's on the other end of a socket, it's oddly satisfying.

    Install

    Ships in the SimpleSyrup pack. ComfyUI Manager → search SimpleSyrup → Install → restart, or:

    Set-Location ComfyUI\custom_nodes
    git clone https://github.com/Artificial-Sweetener/SimpleSyrup.git
    Set-Location SimpleSyrup
    ..\..\venv\Scripts\python.exe -m pip install -r requirements.txt
    

    Restart ComfyUI; current build required (v3 extension API).

    Common issues

    • JSON is empty or says nothing useful - the output only reflects the catalog entries the pack knows; a model you added manually outside the catalog won't have rich metadata.
    • Dropdown options depend on the setting - with "SimpleSyrup: Show downloadable models in loader dropdowns" disabled, only locally-verifiable models appear; that affects the loaders, and this node mirrors the same catalog.
    • Node missing - update ComfyUI first.

    It's the pack's way of saying "here's what you're running, in plain text" - a small, honest utility.

    CategorySimpleSyrup/Masking

    Inputs (2)

    NameTypeDefaultDescription
    sam_modelCOMBOsam_hq_vit_b (379MB)SAM model choice used to create masks from detected boxes.
    grounding_dino_modelCOMBOGroundingDINO_SwinT_OGC (694MB)GroundingDINO model choice used to find boxes that match a text prompt.

    Outputs (1)

    NameTypeDescription
    model_infoSTRINGJSON metadata describing the selected SAM and GroundingDINO model sources.