Edit a video with Gemini Omni Flash on Sume: fields allowed or refused

To edit an existing video with Gemini Omni Flash 1.1 on Sume, send video_url and a prompt that states the edit, through the Video Router. Resolution is optional and defaults to 720p. Do not send aspect_ratio or duration, and do not combine video_url with image_url, end_image_url or any reference_*_urls, because video_url is the edit source, not a reference.

Field rules for edit mode

Taken from the Sume Video Router page, read 2026-10-09. The model ID is gemini-omni-flash-1.1; there is no separate edit model.

A request that follows the rules

The prompt below follows the example in the docs: it names the change and says what to keep. Send an Idempotency-Key so that a retry returns the original job.

curl -X POST https://api.sume.com/v1/video-router/generate \
  -H "Authorization: Bearer $SUME_API_KEY" \
  -H "Content-Type: application/json" \
  -H "Idempotency-Key: edit-clip-001" \
  -d '{
    "model": "gemini-omni-flash-1.1",
    "prompt": "Replace the bottle with an apple. Keep everything else the same.",
    "video_url": "https://example.com/clip.mp4",
    "resolution": "720p",
    "mode": "async"
  }'

Writing the edit prompt

The docs example gives the pattern: say what changes, then say what must stay. 'Replace the bottle with an apple. Keep everything else the same.' is one change and one constraint. Edits that stack several changes in one prompt are harder to check, so run them as separate edits if the first result drifts.

Because edit mode takes no duration or aspect_ratio, you cannot use those fields to reshape the clip. If you want a different ratio, crop or fit the edited clip afterwards with a timeline or filter step rather than adding the field to the edit call, which the docs say is not sent.

The Video Router response is the standard Sume job envelope. Poll the job, or send a webhook URL, and read the result when the status is completed. If the source URL cannot be fetched, the errors page lists input_media_unreachable; make sure the source is a public HTTPS URL that stays valid for the duration of the job.

What it costs

Sume bills Omni by output second and resolution: 360p $0.0375, 720p $0.125, 1080p $0.1875 and 4K $0.375. The edit docs do not say how the output length is chosen, so this page does not assume it. If the output turns out to be as long as a 6-second source, the default 720p edit would be 6 x $0.125 = $0.75, and the usage.cost field on the completed job gives the actual figure.

Google's pricing page lists Omni Flash video output at $17.50 per 1M tokens, about $0.10 per second at 720p; Sume's $0.125 is that figure times 1.25. The reservation at submit is the full estimated price, and a failed edit job releases it.