Categories:

String & binary functions (Large Language Model)

AI_EMBED

Note

AI_EMBED is the updated version of EMBED_TEXT_1024 (SNOWFLAKE.CORTEX). For the latest functionality, use AI_EMBED.

Creates an embedding vector from text or an image. Embeddings are abstract numerical representations of the features of a piece of text or an image that can be used to determine the degree of similarity between pieces of text or images, which can be used for semantic search, clustering, classification, and other tasks.

Region availability

The following table shows the regions where you can use the AI_EMBED function for text and images:

Data type
AWS US West 2
(Oregon)
AWS US East 1
(N. Virginia)
AWS Europe Central 1
(Frankfurt)
AWS Europe West 1
(Ireland)
AWS AP Southeast 2
(Sydney)
AWS AP Northeast 1
(Tokyo)
Azure East US 2
(Virginia)
Azure West Europe
(Netherlands)
AWS
(Cross-Region)
Text

✔

✔

✔

✔

✔

✔

✔

✔

✔
Image

✔

✔

✔

✔

Syntax

AI_EMBED( <model> , <input> )
Copy

Arguments

Required:

model

A string specifying the vector embedding model to be used to generate an embedding.

For text, you can provide the following values:

  • snowflake-arctic-embed-l-v2.0

  • snowflake-arctic-embed-l-v2.0-8k

  • nv-embed-qa-4

  • multilingual-e5-large

  • voyage-multilingual-2

For images, you can provide only the following value:

  • voyage-multimodal-3

Supported models might have different costs.

input

The string or image (as a FILE object) to generate an embedding from. Images must be:

  • In JPEG, WEBP, PNG, or BMP format

  • No more than 10 MB in size

  • No more than 8,000 x 8,000 pixels

Returns

An embedding vector of type VECTOR derived from the input text or image.

Access control requirements

Users must use a role that has the SNOWFLAKE.CORTEX_USER database role. For more information about this privilege, see Required privileges.

Examples

Text example

In this example, a vector embedding is generated for the phrase hello world using the snowflake-arctic-embed-l-v2.0 model:

SELECT AI_EMBED('snowflake-arctic-embed-l-v2.0', 'hello world');
Copy

Image example

In this example, a vector embedding is generated for a staged image using the voyage-multimodal-3 model:

SELECT AI_EMBED('voyage-multimodal-3',
        TO_FILE ('@my_images', 'CITY_WALKING1.PNG'));
Copy

Limitations

  • Snowflake AI functions don’t support dynamic table incremental refresh.

  • Snowflake AI functions don’t work on FILE objects created from files in the following kinds of stages:

    • Internal stages with encryption mode TYPE = 'SNOWFLAKE_FULL'

    • External stages with any customer-side encrypted mode, such as AWS_CSE or AZURE_CSE.

    • User stage

    • Table stage

    • Stage with double-quoted names

Language: English