OpenAI Batch API limits vs Sume bulk runs: 50,000 vs 100

OpenAI's Batch guide allows up to 50,000 requests in one batch file. A Sume Format bulk run takes 1 to 100 items per queue, with concurrency from 1 to 16. A 50,000-line job becomes 500 queues of 100, and there is no 24-hour completion contract to plan around.

OpenAI figures are from its Batch guide; Sume's from Format bulk runs, both read 2026-09-30.

What are the limits side by side?

How do I map a big file onto Sume?

Split the input into groups of at most 100 items, and send each as its own queue with a fresh Idempotency-Key; the docs warn that replaying a spent key returns 202 with the old queue. Each item has the same body as a single Format run. See Format bulk runs: 100 renders.

How do I read results?

There is no queue webhook; communication.webhook_url is per item, and queue progress comes from polling status_url. Queue completed means every item is terminal, not that all succeeded, so branch on counts.failed.

Can I cancel a queue?

Not as a queue: the docs say there is no public list-queues or cancel-queue endpoint. Cancel a child with POST /v1/format-runs/{run_id}/cancel.