# Available detection models

Detection models available on the connected VIS instances.

Endpoint: GET /models
Version: 2.0.0
Security: engineBasic

## Response 200 fields (application/json):

  - `models` (array, required)
    Detection models available on the connected VIS instances.

  - `models.name` (string, required)
    Model name, as passed to the object detector's model or checkpoint_path.

  - `models.type` (string, required)
    Which detector this model works with.
    Enum: "scene", "object", "vlm", "synthetic"

  - `models.custom` (boolean)
    True for a custom checkpoint, false for a built-in variant.

  - `models.checkpoint_path` (string,null)
    Path on the VIS host; pass as the object detector's checkpoint_path.

  - `models.class_set` (string,null)
    Named class set (e.g. coco) when classes is not inlined.

  - `models.classes` (array,null)
    Classes this model can detect, when not named by class_set.

  - `vlm_defaults` (object,null)
    The VLM prompt/schema defaults reported by the VIS instance the default config names, relayed verbatim — VIS owns the shape, and a config UI shows these as the effective standalone-VLM defaults. Null when VIS is unreachable or reports none.

## Response default fields (application/problem+json):

  - `type` (string)
    URI identifying the problem type.

  - `title` (string, required)
    Short, human-readable summary of the problem type.

  - `status` (integer, required)
    The HTTP status code, repeated from the response line.

  - `detail` (string)
    Human-readable explanation specific to this occurrence.

  - `instance` (string)
    URI identifying this specific occurrence of the problem.