Actions21
- Azure OpenAI Actions
- Batch API Actions
- Content Safety Actions
- Document Intelligence Actions
- Language Actions
- Speech Actions
- Translator Actions
- Vision Actions
Batch API → Start Batch
AI-generatedSummary
Upload a JSONL file from a previous node's binary property and start a new Azure OpenAI batch inference job targeting a specified endpoint path, with a maximum completion window currently limited to 24 hours.
Inputs
- Authentication — Choose between API Key or Microsoft Entra ID (OAuth2) for authentication.
- Binary Property (required) — Name of the binary property containing the JSONL file to upload for the batch request.
- Batch Endpoint (required) — Select the Azure OpenAI API endpoint path for the batch job, such as Chat Completions (/chat/completions), Completions (/completions), Embeddings (/embeddings), or specify a custom endpoint path.
- Custom Endpoint Path — Enter a custom relative endpoint path if 'Custom' is selected for Batch Endpoint.
- Completion Window (required) — Set the maximum time window for the batch job to complete. Currently, only '24 Hours' is supported.
Output shape
a single JSON object representing the created batch job response
The operation uploads the JSONL file, creates the batch job, and returns the batch job details including batch job ID and status. Errors are thrown if required parameters are missing or upload fails.
Examples
Example 1: Starting a batch job with a JSONL file stored in a binary property named 'data' targeting the /chat/completions endpoint.
Set Authentication to 'API Key' or 'Entra ID (OAuth2)' as appropriate, Batch Binary Property to 'data', Batch Endpoint to '/chat/completions', and Completion Window to '24 Hours'. The operation uploads the file and starts the batch job, returning the job metadata.