Oracle | DBMS_VECTOR

Oracle DBMS_VECTOR_DATABASE Version 26ai

General Information

Library Note

Morgan's Library Page Header

The best Oracle News for FY2026

Oracle Database 26ai is now availble. If you haven't you downloaded already: Why?

Purpose Provides procedures and functions for vector cloud database service. This service uses a fixed simple schema and convenient way to use that schema. Creation, storage, indexing, search for vector data makes the core capabilities.

AUTHID CURRENT_USER

Constants

Name	Data Type	Value
General
CONST_EXCEP_TABLE_DOES_NOT_EXIST	NUMBER	-942
CONST_EXCEP_TABLE_ALREADY_EXISTS	NUMBER	-955
CONST_EXCEP_INDEX_ALREADY_EXISTS	NUMBER	-955
CONST_EXCEP_INDEX_DOES_NOT_EXIST	NUMBER	-1418
CONST_EXCEP_MODEL_ALREADY_EXISTS	NUMBER	-40204
CONST_EXCEP_MODEL_DOES_NOT_EXIST	NUMBER	-40284
CONST_EXCEP_INVALID_EMBED_PARAMS	NUMBER	-57700
CONST_EXCEP_NO_CLAUSE_IN_QUEREY_BY	NUMBER	-57701
CONST_EXCEP_MULTIPLE_CLAUSES_IN_QUERY_BY	NUMBER	-57702
CONST_EXCEP_PARAMETER_NULL_OR_EMPTY	NUMBER	-57703
CONST_EXCEP_PARAMETER_INVALID_NAME	NUMBER	-57704
CONST_EXCEP_MODEL_NAME_ALREADY_EXISTS	NUMBER	-57705
CONST_EXCEP_METADATA_MISSING	NUMBER	-57706
CONST_EXCEP_FIELDS_MISSING	NUMBER	-57707
CONST_EXCEP_USER_SCHEMA_NOT_SAME	NUMBER	-57708
CONST_EXCEP_UNSUPPORTED_OPERATOR_VALUE	NUMBER	-57709
CONST_EXCEP_UNSUPPORTED_OPERATOR	NUMBER	-57710
CONST_EXCEP_EXCEED_MAX_FILTER_NESTING_DEPTH	NUMBER	-57711
CONST_EXCEP_FILTER_CHECK_BOOLEAN_VALUE	NUMBER	-57712
CONST_EXCEP_FILTER_CHECK_NOT_BOOLEAN_VALUE	NUMBER	-57713
CONST_EXCEP_ONE_KEY_ROOT_LEVEL	NUMBER	-57714
CONST_EXCEP_VALIDATION_ERROR	NUMBER	-57715
CONST_EXCEP_CLOUD_SERVICE_NOT_SET	NUMBER	-57716
CONST_EXCEP_ID_ALREADY_EXISTS_LOAD_VECTORS	NUMBER	-57717
CONST_EXCEP_INDEXING_MANUAL	NUMBER	-57718
CONST_EXCEP_INVALID_INDEXING_VALUE	NUMBER	-57719
CONST_EXCEP_INVALID_EMBED_PATH_DURING_UPSERT	NUMBER	-57720
CONST_EXCEP_BYOV_EMBED_PARAMS	NUMBER	-57721
CONST_EXCEP_INVALID_SEARCH_ID	NUMBER	-57722
CONST_EXCEP_NO_VECTOR_TO_SEARCH	NUMBER	-57723

Data Types -- Types for list_models_pipelined function TYPE list_models_type IS RECORD ( model_name VARCHAR2(256), algorithm VARCHAR2(256), mining_function VARCHAR2(256), creation_date TIMESTAMP WITH TIME ZONE, attributes CLOB); TYPE list_models_type_t IS TABLE OF list_models_type; TYPE list_tables_type IS RECORD ( table_name VARCHAR2(256), description VARCHAR2(4000), auto_generate_id NUMBER, annotations CLOB, vector_type VARCHAR2(256), vector_table_type VARCHAR2(256), embed_params CLOB, index_params CLOB, owner VARCHAR2(256), dense_idx_name VARCHAR2(256), status VARCHAR2(256), stats CLOB, created TIMESTAMP WITH TIME ZONE, updated TIMESTAMP WITH TIME ZONE); TYPE list_vector_tables_type IS TABLE OF list_tables_type;

Dependencies

ALL_ANNOTATIONS_USAGE	DBMS_SQL	JSON_ARRAY_T
ALL_INDEXES	DBMS_STANDARD	JSON_ELEMENT_T
ALL_MINING_MODELS	DBMS_SYSTEM	JSON_KEY_LIST
ALL_MINING_MODEL_ATTRIBUTES	DBMS_SYS_ERROR	JSON_OBJECT_T
ALL_OBJECTS	DBMS_VECTOR	JSON_SCALAR_T
DBMS_ASSERT	DBMS_VECTOR_DATABASE_ADMIN	PLITBLM
DBMS_JSON_SCHEMA	DBMS_VECTOR_LIB	USER_INDEXES
DBMS_LOB	DUAL	USER_TABLES
DBMS_OUTPUT

Documented Yes

Exceptions Exceptions are declared above under CONSTANTS an instantiated using PRAGMA EXCEPTION_INIT

First Available 26ai

Pragma PRAGMA SUPPLEMENTAL_LOG_DATA(DEFAULT, NONE);

Security Model Owned by SYS with EXECUTE granted to PUBLIC.

Source {ORACLE_HOME}/rdbms/admin/dbmsvectordb.sql

Subprograms

CREATE_INDEX	LIST_VECTORS
CREATE_VECTOR_TABLE	LIST_VECTOR_TABLES
CREATE_VECTOR_TABLE_FOR_MODEL	LIST_VECTOR_TABLES_PIPELINED
DELETE_VECTORS	LOAD_MODEL
DESCRIBE_MODEL	LOAD_VECTORS
DESCRIBE_VECTOR_TABLE	PRETTY_PRINT
DROP_INDEX	REBUILD_INDEX
DROP_MODEL	RERANK
DROP_VECTOR_TABLE	SEARCH
GENERATE_EMBEDDING	SUMMARY
INDEX_BUILD_STATUS	UPDATE_VECTOR_TABLE_ANNOTATION
LIST_MODELS	UPSERT_VECTORS
LIST_MODELS_PIPELINED

CREATE_INDEX (new 26ai)

Creates a vector index on the specified vector table column dbms_vector_database.create_index( table_name IN VARCHAR2, index_params IN JSON DEFAULT NULL, debug_flags IN JSON DEFAULT NULL, request_id IN VARCHAR2 DEFAULT NULL) RETURN CLOB;

TBD

CREATE_VECTOR_TABLE (new 26ai)

Creates a "bring-your-own-vector" (BYOV) table with optional metadata and index dbms_vector_database.create_vector_table( table_name IN VARCHAR2, description IN VARCHAR2 DEFAULT NULL, auto_generate_id IN BOOLEAN DEFAULT FALSE, annotations IN JSON DEFAULT NULL, vector_type IN VARCHAR2 DEFAULT 'dense', index_params IN JSON DEFAULT NULL, debug_flags IN JSON DEFAULT NULL, request_id IN VARCHAR2 DEFAULT NULL) return CLOB;

TBD

CREATE_VECTOR_TABLE_FOR_MODEL (new 26ai)

Creates a new "model-integrated" vector table with optional metadata and index dbms_vector_database.create_vector_table_for_model( table_name IN VARCHAR2, description IN VARCHAR2 DEFAULT NULL, auto_generate_id IN BOOLEAN DEFAULT FALSE, annotations IN JSON DEFAULT NULL, embed_params IN JSON, index_params IN JSON DEFAULT NULL, debug_flags IN JSON DEFAULT NULL, request_id IN VARCHAR2 DEFAULT NULL) return CLOB;

TBD

DELETE_VECTORS (new 26ai)

Deletes vectors from the specified vector table

Overload 1 dbms_vector_database.delete_vectors( table_name IN VARCHAR2, ids IN JSON, debug_flags IN JSON DEFAULT NULL, request_id IN VARCHAR2 DEFAULT NULL, output OUT CLOB);

TBD

Overload 2 dbms_vector_database.delete_vectors( table_name IN VARCHAR2, ids IN JSON, debug_flags IN JSON DEFAULT NULL, request_id IN VARCHAR2 DEFAULT NULL) RETURN CLOB;

TBD

DESCRIBE_MODEL (new 26ai)

Retrieves detailed information about a specific mining model, including its name, algorithm, mining function, creation date, and a list of attributes with details such as attribute name, type, data type, length, and vector info. Returns the information as a JSON object dbms_vector_database.describe_model( model_name IN VARCHAR2, debug_flags IN JSON DEFAULT NULL, request_id IN VARCHAR2 DEFAULT NULL) RETURN CLOB;

TBD

DESCRIBE_VECTOR_TABLE (new 26ai)

Retrieves metadata for a vector table in the current schema (integrated model) dbms_vector_database.describe_vector_table( table_name IN VARCHAR2, debug_flags IN JSON DEFAULT NULL, request_id IN VARCHAR2 DEFAULT NULL) RETURN CLOB;

TBD

DROP_INDEX (new 26ai)

Undocumented

Overload 1 dbms_vector_database.drop_index( table_name IN VARCHAR2, debug_flags IN JSON DEFAULT NULL, request_id IN VARCHAR2 DEFAULT NULL, output OUT CLOB);

TBD

Overload 2 dbms_vector_database.drop_index( table_name IN VARCHAR2, debug_flags IN JSON DEFAULT NULL, request_id IN VARCHAR2 DEFAULT NULL) RETURN CLOB;

TBD

DROP_MODEL (new 26ai)

Drops a machine learning or ONNX model from the database environment. After execution, the model and its associated metadata are removed, and a JSON confirmation message is returned dbms_vector_database.drop_model( model_name IN VARCHAR2, debug_flags IN JSON DEFAULT NULL, request_id IN VARCHAR2 DEFAULT NULL) RETURN CLOB;

TBD

DROP_VECTOR_TABLE (new 26ai)

Drops a specified vector table and all its associated vector indexes for the integrated model

Overload 1 dbms_vector_database.drop_vector_table( table_name IN VARCHAR2, debug_flags IN JSON DEFAULT NULL, request_id IN VARCHAR2 DEFAULT NULL, output OUT CLOB);

TBD

Overload 2 dbms_vector_database.drop_vector_table( table_name IN VARCHAR2, debug_flags IN JSON DEFAULT NULL, request_id IN VARCHAR2 DEFAULT NULL) RETURN CLOB;

TBD

GENERATE_EMBEDDING (new 26ai)

Generates vector embeddings for a list of input texts using the specified model dbms_vector_database.generate_embedding( model_name IN VARCHAR2, inputs IN sys.JSON_ARRAY_T, debug_flags IN JSON DEFAULT NULL, request_id IN VARCHAR2 DEFAULT NULL) RETURN CLOB;

TBD

INDEX_BUILD_STATUS (new 26ai)

Retrieves the current build/rebuild status of the VECTOR index associated with the specified vector table dbms_vector_database.index_build_status( table_name IN VARCHAR2, debug_flags IN JSON DEFAULT NULL, request_id IN VARCHAR2 DEFAULT NULL) RETURN CLOB;

TBD

LIST_MODELS (new 26ai)

Retrieves all mining models available in the database along with their attributes dbms_vector_database.list_models( debug_flags IN JSON DEFAULT NULL, request_id IN VARCHAR2 DEFAULT NULL) RETURN CLOB;

TBD

LIST_MODELS_PIPELINED (new 26ai)

Retrieves all mining models available in the database along with their attributes dbms_vector_database.list_models_pipelined RETURN list_models_type_t PIPELINED;

TBD

LIST_VECTORS (new 26ai)

Retrieves a list of vectors and associated metadata from a specified vector table dbms_vector_database.list_vectors( table_name IN VARCHAR2, ids IN JSON DEFAULT NULL, limit IN NUMBER DEFAULT 15, offset IN NUMBER DEFAULT NULL, debug_flags IN JSON DEFAULT NULL, request_id IN VARCHAR2 DEFAULT NULL) RETURN CLOB;

TBD

LIST_VECTOR_TABLES (new 26ai)

Retrieves metadata for all vector tables in the current schema (integrated model) dbms_vector_database.list_vector_tables( debug_flags IN JSON DEFAULT NULL, request_id IN VARCHAR2 DEFAULT NULL) RETURN CLOB;

TBD

LIST_VECTOR_TABLES_PIPELINED (new 26ai)

Retrieves metadata for all vector tables in the current schema (integrated model) dbms_vector_database.list_vector_tables_pipelined RETURN list_vector_tables_type PIPELINED;

TBD

LOAD_MODEL (new 26ai)

Loads a machine learning or ONNX model from a specified URL into the database environment dbms_vector_database.load_model( model_name IN VARCHAR2, url IN VARCHAR2, model_params IN JSON DEFAULT NULL, debug_flags IN JSON DEFAULT NULL, request_id IN VARCHAR2 DEFAULT NULL) RETURN CLOB;

TBD

LOAD_VECTORS (new 26ai)

Loads vector data from a CSV file stored in cloud storage into a database vector table. Depending on the vector_table_type, it may also generate embeddings for the loaded data using a specified model

Overload 1 dbms_vector_database.load_vectors( table_name IN VARCHAR2, url IN VARCHAR2, params IN JSON DEFAULT NULL, debug_flags IN JSON DEFAULT NULL, request_id IN VARCHAR2 DEFAULT NULL, output OUT CLOB);

TBD

Overload 2 dbms_vector_database.load_vectors( table_name IN VARCHAR2, url IN VARCHAR2, params IN JSON DEFAULT NULL, debug_flags IN JSON DEFAULT NULL, request_id IN VARCHAR2 DEFAULT NULL) RETURN CLOB;

TBD

PRETTY_PRINT (new 26ai)

Pretty prints any JSON string in a CLOB variable dbms_vector_database.pretty_print(input_clob IN CLOB);

TBD

REBUILD_INDEX (new 26ai)

Rebuilds the dense vector index for a specified vector table

Overload 1 dbms_vector_database.rebuild_index( table_name IN VARCHAR2, index_params IN JSON DEFAULT NULL, debug_flags IN JSON DEFAULT NULL, request_id IN VARCHAR2 DEFAULT NULL, output OUT CLOB);

TBD

Overload 2 dbms_vector_database.rebuild_index( table_name IN VARCHAR2, index_params IN JSON DEFAULT NULL, debug_flags IN JSON DEFAULT NULL, request_id IN VARCHAR2 DEFAULT NULL) RETURN CLOB;

TBD

RERANK (new 26ai)

Reranks a list of documents against a given query using the specified model dbms_vector_database.rerank( model_name IN VARCHAR2, query IN VARCHAR2, documents IN JSON, model_params IN JSON DEFAULT NULL, debug_flags IN JSON DEFAULT NULL, request_id IN VARCHAR2 DEFAULT NULL) RETURN CLOB;

TBD

SEARCH (new 26ai)

Provides flexible semantic, vector search capabilities over a vector table dbms_vector_database.search( table_name IN VARCHAR2, query_by IN JSON, filters IN JSON DEFAULT NULL, top_k IN NUMBER, include_vectors IN BOOLEAN DEFAULT FALSE, output_selector IN JSON DEFAULT NULL, advanced_options IN JSON DEFAULT NULL, debug_flags IN JSON DEFAULT NULL, request_id IN VARCHAR2 DEFAULT NULL) RETURN CLOB;

TBD

SUMMAARY (new 26ai)

Undocumented dbms_vector_database.summary( debug_flags IN JSON DEFAULT NULL, request_id IN VARCHAR2 DEFAULT NULL) RETURN CLOB;

TBD

UPDATE_VECTOR_TABLE_ANNOTATION (new 26ai)

Updates metadata annotations, description, status, and statistics for an existing integrated model vector table. This function validates that the table exists in the current schema and in the vector database metadata before applying updates

Overload 1 dbms_vector_database.update_vector_table_annotation( table_name IN VARCHAR2, description IN VARCHAR2 DEFAULT NULL, annotations IN JSON DEFAULT NULL, status IN VARCHAR2 DEFAULT NULL, stats IN JSON DEFAULT NULL, debug_flags IN JSON DEFAULT NULL, request_id IN VARCHAR2 DEFAULT NULL, output OUT CLOB);

TBD

Overload 2 dbms_vector_database.update_vector_table_annotation( table_name IN VARCHAR2, description IN VARCHAR2 DEFAULT NULL, annotations IN JSON DEFAULT NULL, status IN VARCHAR2 DEFAULT NULL, stats IN JSON DEFAULT NULL, debug_flags IN JSON DEFAULT NULL, request_id IN VARCHAR2 DEFAULT NULL) RETURN CLOB);

TBD

UPSERT_VECTORS (new 26ai)

Performs an upsert (insert or update) of vectors into a specified vector table

Overload 1 dbms_vector_database.upsert_vectors( table_name IN VARCHAR2, vectors IN JSON, debug_flags IN JSON DEFAULT NULL, request_id IN VARCHAR2 DEFAULT NULL, output OUT CLOB);

TBD

Overload 2 dbms_vector_database.upsert_vectors( table_name IN VARCHAR2, vectors IN JSON, debug_flags IN JSON DEFAULT NULL, request_id IN VARCHAR2 DEFAULT NULL) RETURN CLOB;

TBD