Model DDLs

Applies toBYOCSelf-Managed v3

1 min read

On this page

Usage Notes
CREATE MODEL
View Models
Modify Models
- Examples
Delete Models
- Example

This topic describes the data definition language (DDL) statements for registering, viewing, modifying, and deleting AI models.

Usage Notes

The system supports various large model services with OpenAI compatible interfaces.

CREATE MODEL

Register a Model

Run a CREATE MODEL command in the SQL editor to register a model.

Syntax

SQL

1CREATE [TEMPORARY] MODEL [catalog_name.][db_name.]model_name
2INPUT ( { <physical_column_definition> [, ...n] )
3OUTPUT ( { <physical_column_definition> [, ...n] )
4WITH (key1=val1, key2=val2, ...)
5
6<physical_column_definition>:
7  column_name column_type [COMMENT column_comment]

Clause	Description	Key parameters	Schema requirements	Example
Clause	Description	Key parameters	Schema requirements	Example
INPUT	Defines the fields, field types, and field order of input data.	column_name, column_type, COMMENT	Exactly one STRING field is required.	INPUT (input_text STRING COMMENT 'User comment')
OUTPUT	Defines the fields, field types, and field order of output data.	column_name, column_type, COMMENT	Constraints for different task types:- chat/completions: Exactly one STRING field is required.- embeddings: Exactly one ARRAY<FLOAT> field is required.	OUTPUT (sentiment_label STRING COMMENT 'Sentiment label')
WITH	See Parameters.	provider, endpoint, apiKey, model	None.	WITH ('provider'='openai', 'endpoint'='${ENDPOINT}', 'model'='gpt-4.1', 'apiKey'='${KEY}')

Examples

Completion Model

The following example registers a model for sentiment analysis using an OpenAI-compatible completion service. You can use a proxy to access the API if needed.

SQL

1CREATE MODEL sentiment_analysis_model
2INPUT (input STRING)
3OUTPUT (output STRING)
4WITH (
5  'provider' = 'openai',
6  'task' = 'completions',
7  'endpoint' = 'https://proxy.example.com/v1/chat/completions',
8  'apiKey' = '<your-api-key>',
9  'system_prompt' = 'Analyze the sentiment of the text and return only POSITIVE, NEGATIVE, or NEUTRAL.',
10  'model' = 'gpt-4.1'
11);

When you call this model, the following happens:

You pass a string of text via the input column.
The model sends this to the API with the system prompt.
The model analyzes the text and responds with POSITIVE, NEGATIVE, or NEUTRAL.
The result is returned in the output field.

Embedding Model

The following example registers an embedding model to convert text into numerical vectors.

SQL

1CREATE MODEL text_embedding_model
2INPUT (input STRING)
3OUTPUT (embedding ARRAY<FLOAT>)
4WITH (
5  'provider' = 'openai',
6  'endpoint' = 'https://api.example.com/v1/chat/embeddings',
7  'apiKey' = '<your-api-key>',
8  'model' = 'text-embedding-v3'
9);

These embeddings are useful for:

Semantic search: Find relevant documents based on meaning rather than keyword matches.
Duplicate detection: Identify semantically identical entries.
Text clustering: Automatically group similar documents.
Recommendation systems: Suggest items based on content similarity.
Anomaly detection: Find unusual or unexpected text entries.

Parameters

General

Parameter	Description	Data type	Required	Default value	Note
Parameter	Description	Data type	Required	Default value	Note
provider	The model service provider.	String	Yes	None	Set to openai for OpenAI-compatible services.
endpoint	The API endpoint for chat, completions, or embeddings.	String	Yes	None	Refer to your service's API documentation for the correct endpoint.
api-key	The API key used to access the model service.	String	Yes	None	In VERA Engine 4.1, use apiKey.
max-context-size	The maximum context length in a single request.	Integer	No	None	Supported in VERA Engine 4.2 and later.
context-overflow-action	Specifies how to handle a context exceeding the limit.	String	No	truncated-tail	Supported in VERA Engine 4.2 and later. Valid values: truncated-tail, truncated-tail-log, truncated-head, truncated-head-log, skipped, skipped-log.

chat/completions

Parameter	Description	Data type	Required	Default value
Parameter	Description	Data type	Required	Default value
model	The model to call (e.g., gpt-4.1).	String	Yes	None
system-prompt	The system prompt for the request.	String	Yes	"You are a helpful assistant."
temperature	Controls the randomness of the output (0 to 2).	Float	No	None
top-p	Specifies cumulative probability for sampling.	Float	No	None
stop	The stop sequence string.	String	No	None
max-tokens	Maximum length of generated texts.	Integer	No	None

embeddings

Parameter	Description	Data type	Required	Default value	Note
Parameter	Description	Data type	Required	Default value	Note
model	The model to call.	String	Yes	None
dimension	Controls the output vector dimension.	Integer	No	1024	Valid values: 1024, 768, 512.

View Models

Use the following commands to view information about registered models.

Show registered models:

SQL

1SHOW MODELS [ ( FROM | IN ) [catalog_name.]database_name ];

Show model registration statement:

SQL

1SHOW CREATE MODEL [catalog_name.][db_name.]model_name;

Show model schema:

SQL

1DESCRIBE MODEL [catalog_name.][db_name.]model_name;

Modify Models

Use the ALTER MODEL command to modify an existing model.

SQL

1ALTER MODEL [IF EXISTS] [catalog_name.][db_name.]model_name {
2  RENAME TO new_model_name
3  SET (key1=val1, ...)
4  RESET (key1, ...)
5}

Examples

Rename a model:

SQL

1ALTER MODEL m RENAME TO m1;

Modify a parameter:

SQL

1ALTER MODEL m SET ('endpoint' = 'https://new-endpoint.example.com');

Reset a parameter:

SQL

1ALTER MODEL m RESET ('endpoint');

Delete Models

Use the DROP MODEL command to delete a registered model.

SQL

1DROP [TEMPORARY] MODEL [IF EXISTS] [catalog_name.][db_name.]model_name

Example

SQL

1DROP MODEL sentiment_analysis_model;

Was this helpful?

Yes No