Gemini Omni video edit on Sume: how an unknown clip length is priced

Gemini Omni Flash 1.1 can edit an existing clip: you send video_url and a prompt, and the output length follows the source video. Sume reserves 8 seconds when it does not know the clip length, which is $1.00 at the default 720p ($0.125 per second x 8). You can pass a duration as a length hint of up to 30 seconds, and that hint is used for the reserve estimate only.

This post covers the Video Router edit shape and how the reserve works, as of 2026-10-08. Prices and limits come from the Sume docs and catalog.

What the edit request looks like

Send model: "gemini-omni-flash-1.1", a prompt that gives the instructions, and video_url. resolution is optional and defaults to 720p. Do not send aspect_ratio or duration expecting them to change the output: the catalog says aspect_ratio is rejected for edits and duration is not forwarded to the provider.

video_url is the edit source, not a reference. The docs say you cannot send it together with image_url, end_image_url or reference_*_urls. Sume routes the request by its shape, so you never choose an endpoint: the same model id covers text, image, reference and edit modes.

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: omni-edit-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"}'

Why the hint matters

Sume reserves funds at submit time. If your source clip is 25 seconds long and you send no hint, the estimate uses 8 seconds, which understates the work. A hint of 25 makes the reserve 25 x $0.125 = $3.125 at 720p. Because the hint is not forwarded, it cannot change the output; it only makes the reserve realistic. Sume bills the provider list x 1.25 per output second as a function of the resolution, and usage.cost on the completed job is the billable amount.

Example reserve

For a 12-second source clip, send duration: 12 as a hint. At 720p the reserve is 12 x $0.125 = $1.50; at 1080p it is 12 x $0.1875 = $2.25. The result length still follows the source.

Related limits

• The Video Router lists Gemini Omni Flash 1.1 at 3 to 10 seconds for generation, but the edit hint ceiling is 30 seconds.
• Native synced audio is always on, and the catalog has no audio toggle.
• The edit response is a normal async job: poll GET /v1/jobs/{id}/status or use callback_url on the /v1/videos shape.
• Google describes Gemini Omni Flash as supporting conversational video editing (Google page, read 2026-10-08); Sume exposes the one-shot edit through video_url.