Sync Labs batch API (20 to 500) vs Sume bulk runs (1 to 100)

Sync Labs' Batch API takes 20 to 500 lip sync generations in one JSONL file, targets a 24-hour turnaround, and is limited to Scale and Enterprise plans. Sume has no batch file: bulk runs queue 1 to 100 Format runs in one request, with a concurrency window of 1 to 16 that you poll while it runs.

Sync facts are from its batch guide; Sume facts from Bulk runs, read 2026-10-01.

What are the limits on each side?

What is the unit of work on Sume?

A bulk request is a server-side queue of ordinary Format runs, not a lip sync batch. Each item is one sandbox, one agent turn and one run receipt. Sync's batch guide also says only inputs up to 30 seconds are supported in batch processing.

How do I submit a batch on Sume?

Send concurrency and items with a fresh Idempotency-Key for every batch: replaying a spent key returns 202 with the old queue. One bad item fails the whole create with 400 invalid_request before anything is dispatched. Poll the queue's status_url; completed means every item is terminal, so branch on counts.failed. Similar limits are compared in Gemini batch vs Sume bulk runs.

Is the concurrency window a rate limit?

No. The admission docs call concurrency a dispatch limit, not a submit limit, so queued work waits for a slot instead of being rejected, as long as queue capacity remains.