Vectorize

azure_openai_vectorize

azure_openai_vectorize(batch, api_key=None, azure_endpoint=None, api_version=None, model=None, max_retries=5, provider=None, base_url=None, openai_client=None)

Generates vector embeddings of each string item in a list. Uses the Azure OpenAI service.

Inputs are made safe for the embeddings API before sending: each item is truncated to the model's per-input token limit (a long code prompt or query response would otherwise 400 the request), and the batch is chunked to stay under the endpoint's per-request input cap. One embedding is returned per input string, in the original order, regardless of how many requests the batch is split across.

Note: temperature is intentionally not exposed here — it does not apply to embedding generation, which is deterministic at the API level.

Parameters:

Name	Type	Description	Default
`batch`	`list[str]`	A list of strings to vectorize. Items should be non-empty; the embeddings API rejects empty strings, and items are matched to outputs by position so they are never dropped.	required
`api_key`	`str \| None`	Access key for the Azure OpenAI resource	`None`
`azure_endpoint`	`str \| None`	Your Azure endpoint, including the resource, e.g. https://example-resource.azure.openai.com/. if not provided will default to environment variable `AZURE_OPENAI_ENDPOINT`.	`None`
`api_version`	`str \| None`	API version for Azure resource.	`None`
`model`	`str \| None`	Model deployment name within the Azure resource. If not provided will default to environment variable `AZURE_OPENAI_DEPLOYMENT`.	`None`
`max_retries`	`int`	Maximum number of unsuccessful call attempts to the OpenAI service before returning an error.	`5`
`provider`	`str \| None`	Optional model provider ("azure_openai" default, or "openai_compatible"). Defaults to env `MODEL_PROVIDER`, then Azure OpenAI.	`None`
`base_url`	`str \| None`	Base URL for an OpenAI-compatible endpoint (ignored for Azure).	`None`
`openai_client`	`LanguageModelClient \| None`	Optional pre-built client to reuse. If not provided, one is created for the resolved provider.	`None`

Returns:

Name	Type	Description
`embeddings`	`list[list[float]]`	A list of vector embeddings, one per input string.