Skip to main content

Batch Job

Submit a pre-built JSONL batch file (chat completions or embeddings) to a configured AI provider's batch endpoint and emit the result file when the job finishes. Useful when you have already shaped each input line in the provider's expected format and want the cheapest possible bulk pricing.

For higher-level constructs that build the batch file for you, prefer AI::Enricher, AI::Chunker, or AI::AIEmbeddings.

Pre-requisite: Install an AI provider application (OpenAI, Mistral, Anthropic, …) from Profile > {Organization} > Applications.

Parameters

ProviderREQUIRED

Configured AI application providing the API credentials and provider type.

Model

Model identifier passed to the provider. Defaults to the provider's recommended model when left empty.

EndpointREQUIRED

Which provider endpoint the batch job targets: - Chat completion — each batch line is sent as a chat completion request. - Embeddings — each batch line is sent as an embeddings request.

Input

FileREQUIRED

JSONL batch input file in the provider's expected format (one request per line).

Output

File

JSONL result file returned by the provider when the batch job completes.