openai-batch-helper documentation

Batching can cut OpenAI usage costs by roughly 50%, but juggling JSONL files, uploads, and polling is tedious. This guide shows how openai-batch-helper keeps the batch flow tidy, typed, and production-friendly.

Contents

Guides

• Overview
  • What is openai-batch-helper?
  • Why batch, why this helper?
  • What you get
  • How it works (high level)
  • Public API (library)
  • CLI overview
  • Design principles
  • Non-goals (for now)
• Installation
  • Requirements
  • From PyPI (future)
  • From source (development)
  • Quality checks
• Usage
  • Quickstart (chat with add_task)
  • Add embeddings tasks
  • Progress logging
  • Result parsing
  • Custom extractors
  • Errors file
• Command-line interface
  • Flags
  • Exit codes
• Batch API Background
  • Request JSONL format
  • Lifecycle
  • Outputs
  • Endpoint alignment
• FAQ
  • What does this project do in one sentence?
  • Does it stream responses?
  • Which endpoints are supported?
  • How do I see progress while waiting?
  • What errors can be raised?
  • What Python versions are supported?
  • Can I customize result parsing?

Reference

• API Reference
  • BatchHelper
  • BatchJob
  • status_progress_logger()
  • status_progress_printer()
• Exceptions
  • BatchHelperError
  • BatchNotCompletedError
  • EmptyBatchError

Project links

• Source code: https://github.com/hesenp/openai_batch_helper

• README: https://github.com/hesenp/openai_batch_helper/blob/main/README.md