To provide feedback or request support for this feature, send email to

The INFORMATION_SCHEMA.VECTOR_INDEX_COLUMNS view contains one row for each vector-indexed column on each table in a dataset.

Required permissions

To see vector index metadata, you need the bigquery.tables.get or bigquery.tables.list Identity and Access Management (IAM) permission on the table with the index. Each of the following predefined IAM roles includes at least one of these permissions:

  • roles/bigquery.admin
  • roles/bigquery.dataEditor
  • roles/bigquery.dataOwner
  • roles/bigquery.dataViewer
  • roles/bigquery.metadataViewer
  • roles/bigquery.user

For more information about BigQuery permissions, see Access control with IAM.


When you query the INFORMATION_SCHEMA.VECTOR_INDEX_COLUMNS view, the query results contain one row for each indexed column on each table in a dataset.

The INFORMATION_SCHEMA.VECTOR_INDEX_COLUMNS view has the following schema:

Column name Data type Value
index_catalog STRING The name of the project that contains the dataset.
index_schema STRING The name of the dataset that contains the vector index.
table_name STRING The name of the table that the vector index is created on.
index_name STRING The name of the vector index.
index_column_name STRING The name of the indexed column.
index_field_path STRING The full path of the expanded indexed field, starting with the column name. Fields are separated by a period.

Scope and syntax

Queries against this view must have a dataset qualifier. The following table explains the region scope for this view:

View Name Resource scope Region scope

Replace the following:

  • Optional: PROJECT_ID: the ID of your Google Cloud project. If not specified, the default project is used.
  • DATASET_ID: the ID of your dataset. For more information, see Dataset qualifier.


-- Returns metadata for vector indexes in a single dataset.


The following query extracts information on columns that have vector indexes:

SELECT table_name, index_name, index_column_name, index_field_path

The result is similar to the following:

| table_name | index_name | index_column_name | index_field_path |
| table1     | indexa     | a                 | a                |
| table2     | indexb     | b                 | b                |
| table3     | indexc     | c                 | c                |