Sora remix and edits ended: edit a video with Gemini Omni Flash

To edit an existing video on Sume, call POST /v1/video-router/generate with model set to gemini-omni-flash-1.1, a video_url and a prompt that describes the change. That is the closest match to what OpenAI's guide called Sora edits: targeted changes to an existing video that preserve its structure.

The same guide marked remix as being deprecated in favor of edits, and the whole Videos API was shut down on 2026-09-24 (read 2026-10-01). Sume's side is described in the Video Router docs.

What does an edit request look like?

video_url is the edit source, not a reference, so it cannot be combined with image_url, end_image_url or any reference_*_urls. Resolution is optional and defaults to 720p. aspect_ratio is rejected and no duration is sent to the model.

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"
  }'

What does an edit cost?

Gemini Omni Flash 1.1 bills per output second by resolution at list x 1.25. Because the model takes no duration on an edit, the reserve uses your duration hint when you send one and otherwise assumes 8 seconds. At 720p that reserves $1.00 for 8 seconds.

Which Sora features have no equivalent?

Sora's edits took the earlier video's structure and changed parts of it; Sume's edit is a single prompt over one source video, with no extra references and no choice of aspect ratio or duration. Sume's other video-input models take a video as a reference or for motion transfer rather than prompt edits. Check each result before you rely on it in a pipeline; for prompt wording see Gemini Omni video edit prompts.