Batch Request Builder

Turn a table of rows into a batch job, with an id per request that survives reordering.

Templates and rows — usually real customer data — are expanded in the browser and never uploaded.

Estimate what the job will cost

About Batch Request Builder

Batch processing is the cheapest way to run one prompt over a lot of rows, at about half the interactive price, and the file it wants is unforgiving in one specific way: results come back in whatever order they finish. The custom id is the only thing tying a result to the row it came from, so a duplicate id means two results you cannot tell apart — silently, after the job has run. This expands your template over the rows, derives an id from a column or generates one, renames duplicates rather than letting them collide, and sizes the job so you know what you are submitting before you submit it.

Features

• One request per row, from a template and a table
• Custom id from a column of your choosing, or generated
• Duplicate ids renamed rather than left to collide
• Ids sanitised to the characters the field allows
• Warns per row when a placeholder had no value
• Job size: request count, estimated input tokens and file size
• Emits both the request body and the same requests as JSONL
• Splits an oversized job into files, and shows how to submit it

How to use the Batch Request Builder

1. Paste the prompt template and the rows
2. Pick the model and the per-request token limit
3. Choose which column becomes the custom id
4. Download the request body and submit it

Example

Input

"custom_id": "Acme-Ltd"

Output

Row 2 repeats the id “Acme-Ltd”; renamed to “Acme-Ltd-2”

Results arrive out of order, so the id is the only way back to the row.

Common errors & troubleshooting

• I cannot match results back to rows. — Key on the custom id, never on position — results come back in completion order. Duplicates are renamed here for exactly this reason.
• The submission is rejected. — The endpoint takes a JSON body with a requests array, not the JSONL file. Both are produced; submit the .json one.
• A prompt contains a literal placeholder. — That row had no value for it. Every affected row is listed, with the placeholder named.
• The job is too large. — Split it. Files of at most ten thousand requests are offered automatically when the count goes above that.

Frequently asked questions

Why is batch processing cheaper?

Because it is not interactive. The work is queued and run when there is capacity, which is a much easier thing to schedule — so it is priced at about half. The trade is latency, not quality.

What is the custom id for?

Matching results to inputs. Results arrive in completion order rather than submission order, so the id is the only reliable link back to the row a result came from.

What happens if two rows share an id?

In a real job, two results you cannot tell apart. Here the duplicate is renamed and flagged, which is a much better time to find out.

What is the JSONL output for, if the endpoint takes JSON?

Line-oriented tooling: splitting a large job, diffing two versions of one, or feeding a different runner. The JSON body is the one to submit.

How should I choose max_tokens?

High enough that no answer is truncated, since a truncated result has to be redone. The ceiling shown assumes every request uses its full allowance, which is a worst case rather than an estimate.

Related tools

• Prompt Template Filler — Expand one prompt over a table of rows and see what is missing.
• JSONL Viewer & Validator — Read, filter and check a JSONL file line by line.
• Agent Loop Cost Estimator — Cost an agent loop turn by turn, with history re-sent and caching applied.
• Fine-Tuning Dataset Validator — Check a chat JSONL dataset for the errors that fail a training run.
• Prompt Cache Planner — Order the blocks of a request so the cacheable prefix is as long as possible.
• Context Window Planner — Budget a context window across prompt, tools, history and the reply.

More AI & Agents tools

• Skill Builder — Write a SKILL.md with the frontmatter checked as you type.
• Skill Linter — Paste a SKILL.md and see what stops it being matched to a task.
• Subagent Builder — Build a subagent definition file with its tools and model set.
• Slash Command Builder — Build a custom slash command, with arguments and a tool allow-list.
• Project Instructions Builder — Draft a CLAUDE.md that earns its place in every session.
• Plugin Manifest Builder — Generate a plugin.json and see the folder layout it expects.
• Hooks Builder — Build settings.json hooks and catch the matchers that never fire.
• Permission Rule Tester — Test allow, ask and deny rules against real commands before you trust them.
• MCP Server Config Builder — Generate an MCP server entry for stdio, HTTP or SSE — and the CLI one-liner.
• MCP Tool Schema Builder — Define a tool and get the JSON Schema, with strict mode checked.
• Tool Schema Checker — Paste a tool definition and find out why it is rejected or mis-called.
• Prompt Template Filler — Expand one prompt over a table of rows and see what is missing.
• System Prompt Linter — Find contradictions, dead instructions and unfilled placeholders in a prompt.
• Prompt Diff — Compare two prompt versions word by word, with the token delta.
• Few-Shot Example Builder — Turn input/output pairs into tagged, JSON or chat-format examples.
• XML Prompt Section Builder — Wrap prompt sections in tags that content cannot break out of.
• JSONL Viewer & Validator — Read, filter and check a JSONL file line by line.
• Chat Transcript Converter — Turn a messages array into a readable transcript, and back again.
• Fine-Tuning Dataset Validator — Check a chat JSONL dataset for the errors that fail a training run.
• SSE Stream Viewer — Paste a raw streaming response and get the text, tool calls and usage back.
• Context Window Planner — Budget a context window across prompt, tools, history and the reply.
• Prompt Cache Planner — Order the blocks of a request so the cacheable prefix is as long as possible.
• Text Chunker — Split a document into overlapping chunks that break on real boundaries.
• Agent Loop Cost Estimator — Cost an agent loop turn by turn, with history re-sent and caching applied.

See all AI & Agents tools · All tools