Which Sume Format for a UGC-style ad? Read io and showcase first

For a UGC-style ad, read the candidate Formats before you run any of them: GET /v1/formats/sume/{slug} returns each Format's description, its io profile (input_kind and output_kind) and a showcase that is a real earlier output. Sume's catalog lists sume-close-camera-ugc, sume-mobile-app-ugc, sume-product-usage-demo, sume-before-after, sume-video-hook and sume-virtual-try-on among others, and any key with formats:write can call them.

What the docs say you can rely on

The Format catalog page says the catalog answers at the reserved sume handle, and that you can call each listed slug with POST /v1/formats/sume/{slug}/runs. Any other slug returns 404 format_not_found. The page does not describe what each Format produces in prose, which is why the describing fields matter.

The io profile tells you the input shape: input_kind is url, text, image or product, and output_kind is video, image or text. The showcase is a worked example that the Format really produced; Sume checks it against the generated-media ledger before storing it. Both are null for Formats saved before registration existed, which means not declared, not takes no input.

A selection routine

List the five or six UGC-flavored slugs, GET each one, and compare the showcase videos with your idea. Check that your input matches input_kind: a product page needs a Format whose input_kind is url or product, while a still photo needs image. Choosing by input shape avoids the case where you have a photo and the Format wants a product URL.

Then run one of them with a low spend cap. generation_spend_cap_usd on the request names this run's ceiling, accepted up to $500, and a run that tries to spend more ends as failed with format_run_failed. Compare usage.billable_amount_usd_micros with the cap before you raise the brief.

curl -sS https://api.sume.com/v1/formats/sume/sume-close-camera-ugc \
  -H "Authorization: Bearer $SUME_API_KEY" | head -c 1200

What not to assume

The catalog slugs and their showcases are the evidence; a slug's name is not a spec. Sume does not publish a per-Format price in the docs, because a run is metered at the model rates it uses. Use the cap to bound it and read the receipt after.

Run results belong to the key that made the call, and the catalog Format itself stays shared. To change one, fork it in the Format library and call your copy at your own handle.

A practical note

Use the showcase as your acceptance test. Before running a Format on your product, write down what the showcase does well, such as framing, pacing or voice, and what it does not. After your run, compare your output with the showcase and with your own brief. If your output differs from both, change the input rather than the Format. Bulk runs queue many calls to the same Format, so once one run is accepted, scale it with the bulk-runs page, keeping the spend cap on every item.

Before you scale

Every price in this post is a Sume list price read from the public catalog on 2026-10-09, and the arithmetic is shown so you can redo it with your own counts. Rates can change, so re-read the catalog before a large batch and run a small test first. Sume bills the job's captured amount, and the job result tells you what was used, so compare the first run's receipt with your estimate before you scale the work to the full set.