Skip to main content

​
module  pixeltable.functions.fabric

Pixeltable UDFs that wrap Azure OpenAI endpoints via Microsoft Fabric. These functions provide seamless access to Azure OpenAI models within Microsoft Fabric notebook environments. Authentication and endpoint discovery are handled automatically using Fabric’s built-in service discovery and token utilities. Note: These functions only work within Microsoft Fabric notebook environments. For more information on Fabric AI services, see: https://learn.microsoft.com/en-us/fabric/data-science/ai-services/ai-services-overview

​
udf  chat_completions()

Signature
@pxt.udf
chat_completions(
    messages: pxt.Json,
    *,
    model: pxt.String,
    api_version: pxt.String | None = None,
    model_kwargs: pxt.Json | None = None
) -> pxt.Json
Creates a model response for the given chat conversation using Azure OpenAI in Fabric. Equivalent to the Azure OpenAI chat/completions API endpoint. For additional details, see: https://learn.microsoft.com/en-us/azure/ai-services/openai/reference Automatic authentication: Authentication is handled automatically in Fabric notebooks using token-based authentication. No API keys are required. Supported models in Fabric:
  • gpt-5 (reasoning model)
  • gpt-4.1
  • gpt-4.1-mini
Request throttling: Applies the rate limit set in the config (section fabric.rate_limits, key chat). If no rate limit is configured, uses a default of 600 RPM. Requirements:
  • Microsoft Fabric notebook environment
  • synapse-ml-fabric package (pre-installed in Fabric)
Parameters:
  • messages (pxt.Json): A list of message dicts with ‘role’ and ‘content’ keys, as described in the Azure OpenAI API documentation.
  • model (pxt.String): The deployment name to use (e.g., ‘gpt-5’, ‘gpt-4.1’, ‘gpt-4.1-mini’).
  • api_version (pxt.String | None): Optional API version override. If not specified, defaults to ‘2025-04-01-preview’ for reasoning models (gpt-5) and ‘2024-02-15-preview’ for standard models.
  • model_kwargs (pxt.Json | None): Additional keyword args for the Azure OpenAI chat/completions API. For details on available parameters, see: https://learn.microsoft.com/en-us/azure/ai-services/openai/reference Note: Reasoning models (gpt-5) use max_completion_tokens instead of max_tokens and do not support the temperature parameter.
Returns:
  • pxt.Json: A dictionary containing the response and other metadata.
Examples: Add a computed column that applies the model gpt-4.1 to an existing Pixeltable column tbl.prompt of the table tbl: 
from pixeltable.functions import fabric

messages = [
    {'role': 'system', 'content': 'You are a helpful assistant.'},
    {'role': 'user', 'content': tbl.prompt},
]
tbl.add_computed_column(
    response=fabric.chat_completions(messages, model='gpt-4.1')
)
Using a reasoning model (gpt-5): 
tbl.add_computed_column(
    reasoning_response=fabric.chat_completions(
        messages,
        model='gpt-5',
        model_kwargs={'max_completion_tokens': 5000},
    )
)

​
udf  embeddings()

Signature
@pxt.udf
embeddings(
    input: pxt.String,
    *,
    model: pxt.String = 'text-embedding-ada-002',
    api_version: pxt.String = '2024-02-15-preview',
    model_kwargs: pxt.Json | None = None
) -> pxt.Array[(None,), float32]
Creates an embedding vector representing the input text using Azure OpenAI in Fabric. Equivalent to the Azure OpenAI embeddings API endpoint. For additional details, see: https://learn.microsoft.com/en-us/azure/ai-services/openai/reference Automatic authentication: Authentication is handled automatically in Fabric notebooks using token-based authentication. No API keys are required. Supported models in Fabric:
  • text-embedding-ada-002
  • text-embedding-3-small
  • text-embedding-3-large
Request throttling: Applies the rate limit set in the config (section fabric.rate_limits, key embeddings). If no rate limit is configured, uses a default of 600 RPM. Batches up to 32 inputs per request for efficiency. Requirements:
  • Microsoft Fabric notebook environment
  • synapse-ml-fabric package (pre-installed in Fabric)
Parameters:
  • input (pxt.String): The text to embed (automatically batched).
  • model (pxt.String): The embedding model deployment name (default: ‘text-embedding-ada-002’).
  • api_version (pxt.String): The API version to use (default: ‘2024-02-15-preview’).
  • model_kwargs (pxt.Json | None): Additional keyword args for the Azure OpenAI embeddings API. For details on available parameters, see: https://learn.microsoft.com/en-us/azure/ai-services/openai/reference
Returns:
  • pxt.Array[(None,), float32]: An array representing the embedding vector for the input text.
Examples: Add a computed column that applies the model text-embedding-ada-002 to an existing Pixeltable column tbl.text of the table tbl: 
from pixeltable.functions import fabric

tbl.add_computed_column(embed=fabric.embeddings(tbl.text))
Add an embedding index to an existing column text: 
tbl.add_embedding_index(
    'text',
    embedding=fabric.embeddings.using(model='text-embedding-ada-002'),
)
Last modified on March 1, 2026