Ideogram image URLs expire: what to keep vs Sume job results

Download the image bytes the moment a call succeeds and store the file, not the URL. Ideogram's quickstart says "Image endpoints return results directly; image URLs expire, so download anything you want to keep." The same habit is safe with Sume, whose image results come back as data[].url, a Sume-hosted signed URL.

Ideogram facts are from its developer docs landing page; Sume facts from the Image API docs and Jobs and results, read 2026-10-01.

What should I save after an Ideogram call?

Save the image file itself, plus whatever you need to reproduce or audit it: your prompt, the parameters you sent, and your own record id. The quickstart gives no lifetime for the URL, so treat any returned link as temporary and do not schedule work around a guessed expiry.

What does a Sume image result give me?

The Image API docs describe the result payload as data[].url (Sume-hosted, signed) rather than inline base64. They also say usage.cost is the billed USD amount. Read both from the result, then copy the file if your product needs it long term.

What if the request takes longer than the wait?

Sume caps a blocking wait at 30s. Slower generations, and requests with mode: "async" or "webhook", return the job envelope with 202 and polling URLs; you read the images from the job result endpoint once it is ready. If the wait ends without a terminal job, poll; do not resubmit the paid request. See also Flux image URLs expire in 10 minutes for the same pattern on another model.