Batch image generation API: how many images can run at once?

How many images run at once depends on your plan: Sume's docs list 1 processing job on Free, 4 on Pro, 8 on Startup and 20 on Scale, and valid extra jobs wait as queued instead of failing. Only when the queue is also full does a submit return 429 queue_full.

From the Generation admission and Image API docs, read 2026-09-29. The dashboard Concurrency tab and the generation_limits field in submit responses are the live source, and the docs say to prefer them over the static table.

What are the limits per plan?

What happens when I submit more than the limit?

Valid jobs are accepted as queued while queue capacity remains, then move to processing as slots open. queued is not a failure. When both are full the submit fails with 429 queue_full; wait for jobs to finish or cancel queued ones, then retry with the same idempotency key.

How should I submit a batch?

Use mode: "async" with an Idempotency-Key per item, store each job id, and poll with backoff. Do not resubmit a paid request just because your local worker timed out.

for i in 1 2 3 4; do
  curl -s -X POST https://api.sume.com/v1/images \
    -H "Authorization: Bearer $SUME_API_KEY" \
    -H "Content-Type: application/json" \
    -H "Idempotency-Key: catalog-batch-001-item-$i" \
    -d "{\"model\": \"google/nano-banana-2\", \"prompt\": \"Studio photo, item $i\", \"mode\": \"async\"}"
done

Does a queued job cost anything while it waits?

A submit needs the estimate reserved from your balance, and without enough balance it returns 402 insufficient_credits before any provider work starts. A job that fails or is canceled is not billed for images. Check GET /v1/balance before a large batch.