MaxCompute:AI_GENERATE

Command format

The AI_GENERATE function has different signatures for calling a large language model (LLM) and a multimodal large language model (MLLM).

• When model_type is LLM, the function signature is as follows:

  STRING AI_GENERATE(
    STRING <model_name> , 
    STRING <version_name>, 
    STRING <prompt> 
    [, STRING <model_parameters>]
  );

• When model_type is MLLM, the function signature is as follows:

  STRING AI_GENERATE(
    STRING <model_name>, 
    STRING <version_name>, 
    STRING | BINARY <unstructured_data> ,
    STRING <prompt> 
    [, STRING <model_parameters>]
  );

Parameters

• model_name: Required. A STRING that specifies the name of the model to use. The model can be an LLM or an MLLM.

• version_name: Required. A STRING that specifies the name of the model version to use. To use the default version, set this parameter to DEFAULT_VERSION.

• prompt: Required. A STRING that specifies the prompt to send to the model. The value can be a STRING constant, a column name, or an expression.

• unstructured_data: Required when model_type is MLLM. A STRING or BINARY value that specifies the multimodal data to process. You can specify an image URL as a STRING or the image's binary data as BINARY. If you use the BINARY data type, a BINARY input parameter must also be specified when the model is created.

• model_parameters: Optional. A STRING. You can use this parameter to specify parameters for the model call, such as max_tokens, temperature, and top_p, in the following format: '{"max_tokens": 500, "temperature": 0.6, "top_p": 0.95}'. The parameters are described as follows:

  • max_tokens: The maximum number of tokens in the output from a single model call. For public models in MaxCompute, the default value is 4096.

  • temperature: A value between 0 and 1 that controls the randomness of the model's output. A higher value makes the output more diverse and random.

  • top_p: A value between 0 and 1 that controls the randomness and diversity of the model's output. A higher value makes the output more random.

Return value

Returns a STRING that contains the content generated by the model.

Examples

• Example 1: Call a model for content generation.

  Call the Qwen3-0.6B-GGUF public model in MaxCompute to generate content.

  SET odps.sql.using.public.model=true;
  -- The following result is returned.
  -- "The capital of China is **Beijing**."
  SELECT AI_GENERATE(bigdata_public_modelset.default.Qwen3-0.6B-GGUF,DEFAULT_VERSION,'what is the capital of China');

• Example 2: Call a model for sentiment analysis.

  Call the Qwen3-1.7B-GGUF public model in MaxCompute to perform sentiment analysis on user comments.

  SET odps.sql.using.public.model=true;
  SELECT 
      prompt,
      AI_GENERATE(
          bigdata_public_modelset.default.Qwen3-1.7B-GGUF,
          DEFAULT_VERSION,
          concat('Perform sentiment analysis and classification on the following comment. The output must be one of these three options: Positive, Negative, or Neutral. Comment to analyze:', prompt)
      ) AS generated_text
  FROM (
      VALUES 
          ('The weather is great today, and I''m in a good mood! It''s sunny and perfect for a walk. /no_think'),
          ('The weather is great today, and I''m in a good mood! It''s sunny. /no_think'),
          ('Technology is advancing rapidly, and artificial intelligence is changing lives. /no_think'),
          ('The control measures are excellent. Thumbs up to the medical staff! /no_think'),
          ('The quality of this product is very poor. /no_think')
  ) t (prompt);
  
  -- The following result is returned:
  +------------------------------------------------------------------------------------------------+----------------+
  | prompt                                                                                         | generated_text |
  +------------------------------------------------------------------------------------------------+----------------+
  | The weather is great today, and I'm in a good mood! It's sunny and perfect for a walk. /no_think | "Positive"     |
  | The weather is great today, and I'm in a good mood! It's sunny. /no_think                      | "Positive"     |
  | Technology is advancing rapidly, and artificial intelligence is changing lives. /no_think        | "Positive"     |
  | The control measures are excellent. Thumbs up to the medical staff! /no_think                  | "Positive"     |
  | The quality of this product is very poor. /no_think                                            | "Negative"     |
  +------------------------------------------------------------------------------------------------+----------------+
  

• Example 3: Call a model for multimodal data processing.

  This example calls a model named PAI_EAS_Qwen25_Omni_3B, which is a PAI-EAS remote model, to tag product categories based on an image URL and a prompt. You must also create an Object Table named image_demo. For more information, see Automatically generate E-commerce product descriptions using a MaxCompute remote model.

  Note

  MaxCompute remote models currently support only PAI-EAS Internet addresses. When you use an AI function to call a model, you must add the PAI-EAS Internet address as an available external network address in MaxCompute. For more information, see Network enablement process.

  SELECT
    key,
    AI_GENERATE(
      PAI_EAS_Qwen25_Omni_3B, v1, image_url,
      "Recognize and extract the product category from the e-commerce product sales poster. The result must be one of the following six options: Cosmetics, Apparel, Daily Necessities, Food, Other, Electronic Products. Do not include any other text or information."
    ) AS item_catagory
    FROM (
      SELECT GET_SIGNED_URL_FROM_OSS(
        'pd_test_model.default.image_demo', key, 604800
      ) AS image_url, key AS key
      FROM pd_test_model.default.image_demo
  ) Limit 10;
  
  -- The following result is returned:
  +--------------------+---------------------+
  | key                | item_catagory       |
  +--------------------+---------------------+
  | alimamazszw-1.jpg  | Food                | 
  | alimamazszw-10.jpg | Electronic Products | 
  | alimamazszw-11.jpg | Electronic Products | 
  | alimamazszw-12.jpg | Cosmetics           | 
  | alimamazszw-13.jpg | Electronic Products | 
  | alimamazszw-14.jpg | Daily Necessities   | 
  | alimamazszw-15.jpg | Cosmetics           | 
  | alimamazszw-16.jpg | Cosmetics           | 
  | alimamazszw-18.jpg | Daily Necessities   | 
  +--------------------+---------------------+