# VectorCollection

Model ID: `VectorCollection`

One searchable collection and the facts that decide whether it is usable.

Type: `object`

## Fields

| Field | Type | Required | Format | Allowed values | Default | Nullable | Description |
|---|---|---:|---|---|---|---:|---|
| `available` | `boolean` | Yes |  |  |  | No | Whether the collection could be described and counted. |
| `chunk_count` | `integer` | Yes |  |  |  | No | Number of chunk rows currently loaded. |
| `collection` | `string` | Yes |  |  |  | No | Collection name. |
| `detail` | `string|null` | No |  |  |  | Yes | Why the collection is unavailable, when it is. |
| `dimension` | `integer|null` | No |  |  |  | Yes | Vector width. Must equal the query embedding's width or search cannot work. |
| `metric_type` | `string|null` | No |  |  |  | Yes | Similarity metric the vector index was built with, e.g. COSINE. |
| `partitioned_by_payer` | `boolean` | Yes |  |  |  | No | When true, the collection is grouped by payer and every search of it must supply payer_key. |
| `role` | `string` | Yes |  |  |  | No | Which corpus this collection holds: documents \| policies. |

## Example

None declared in canonical OpenAPI.