Creates a new batch job for processing multiple requests asynchronously.
The model provider (e.g., openai, anthropic, azure, bedrock, gemini)
Model identifier for batch routing
ID of the uploaded JSONL file containing batch requests (OpenAI/Azure style)
Array of inline request items (Anthropic style)
Target endpoint for batch requests
/v1/chat/completions, /v1/embeddings, /v1/completions, /v1/messages Time window for completion (e.g., 24h)
User-provided metadata
Batch created successfully
Unique batch ID
Object type (batch)
Target endpoint
Input file ID
Completion window
Batch status
validating, failed, in_progress, finalizing, completed, expired, cancelling, cancelled Unix timestamp of creation
Unix timestamp of expiration