Batch Prompt Formatter

Apply a prompt template with {column} placeholders across a CSV. Generates one prompt per row, then exports as the batch-format JSONL that OpenAI's and Anthropic's Batch APIs expect. Useful for running 10,000 evaluations or transformations at 50% discounted batch pricing.

How to use the Batch Prompt Formatter

Paste CSV with a header row. Write a prompt template referencing column names as {column}. Each row becomes one batch entry. Output is the JSONL format OpenAI's Batch API accepts (or Anthropic's). Upload to /v1/files, then create a batch job — half-price processing within 24h.

Running one prompt across many rows

When you need to run the same prompt over thousands of inputs — classify every row, translate a column, score a dataset — sending one synchronous request per item is slow and pays full price. Both OpenAI and Anthropic offer a Batch API that takes a JSONL file of requests, processes them asynchronously within 24 hours, and charges about half the normal rate.

This formatter takes a prompt template with {column} placeholders and a CSV, fills the template once per row, and writes the exact JSONL the Batch API expects — each line a request object with its own custom ID. You pick the OpenAI or Anthropic layout and the target model, download the file, and submit it.

Common use cases

• Bulk classification — label every row of a dataset with one prompt run at batch pricing.
• Dataset generation — produce synthetic examples from a template across many seed values.
• Evaluations — run a test prompt over thousands of cases in a single job.
• Content transforms — rewrite, summarize, or translate a column for every record.
• Cost control — trade immediacy for the roughly 50% batch discount on large jobs.