Zapier Next Gen Zaps drop the 500-item loop cap: pace Sume

With no 500-item cap, a Zap can loop over thousands of rows, and the limit that bites is Sume's, not Zapier's. A loop of paid submits can use up accepted-job capacity long before its write budget, so size each wave from your plan.

Zapier facts are from its September 23 announcement; Sume facts from Authentication and Generation admission, read 2026-10-01. Zapier's post gives no concurrency details for loops.

What did Zapier change?

The announcement for Next Gen Zaps lists no more 500-item loop cap, nested loops, branches that split and rejoin, pausing mid-run for a human reply, error handling that keeps going and self-healing monitoring agents. It is in early access on paid plans: Pro, Team and Enterprise.

How many submits will Sume accept?

Two limits apply. Writes per minute are a request budget, and a 429 rate_limited means it is spent. Accepted capacity is processing concurrency plus queue; when it is full, new paid submits fail with 429 queue_full. Queued is a normal state, not a failure.

How big should a wave be?

Read generation_limits on the submit response. Use max(0, concurrency_limit - active_generation_jobs - queued_generation_jobs), capped by queue_capacity_remaining, as the budget for new in-flight work, and refresh before the next wave. The wave_size_hint field is only a hint about submission, never a concurrency setting. A Pro workspace with 24 accepted jobs cannot take a 500-item loop in one go.

What should each loop item send?

A submit with mode: "webhook", a public HTTPS webhook_url, and an Idempotency-Key built from the row id so a Zapier retry returns the original job.

Use a path-based limit in Zapier or a delay step between waves. Zapier's documentation excerpt gives no figure for either, so pace by Sume's numbers.

curl -X POST https://api.sume.com/v1/video-1.0/generate \
  -H "x-api-key: $SUME_API_KEY" \
  -H "content-type: application/json" \
  -H "idempotency-key: row-8823-r1" \
  -d '{"prompt":"Slow push-in on a ceramic mug","mode":"webhook","webhook_url":"https://example.com/hooks/sume"}'

What happens when the queue is full?

Wait for jobs to finish or cancel queued ones, then retry. See queue_full versus a full concurrency cap and Zapier's own webhook limits.