Gemini Omni edit mode on Sume: video_url only, 720p, no aspect_ratio
Omni edit mode takes a prompt and one video_url. Resolution defaults to 720p, aspect_ratio is rejected, and no image or reference field may be added.

To edit an existing clip with gemini-omni-flash-1.1 on Sume, send a prompt that describes the change plus one video_url to POST /v1/video-router/generate. Resolution is optional and defaults to 720p, aspect_ratio is rejected, and you cannot add image_url, end_image_url or any reference_*_urls in the same request. Sume bills the provider list times 1.25 per output second, by resolution.
Edit versus reference-to-video
The two modes use similar media but different fields, and mixing them is the commonest mistake. In edit mode the video is the thing you change. In reference-to-video it is guidance for a new clip.
| Mode | Send | Do not send |
|---|---|---|
| Text-to-video | prompt | media |
| Image-to-video | image_url, optional end_image_url | video_url |
| Reference-to-video | reference_image_urls (up to 10), reference_video_urls (up to 3, each at most 3 s) | video_url |
| Edit | prompt, video_url, optional resolution | aspect_ratio, image fields, reference fields |
Writing the edit prompt
In edit mode the prompt gives the instruction. The docs example is short and states what stays the same: replace the bottle with an apple and keep everything else the same. Naming the unchanged parts helps the model leave them alone.
Duration is not sent in edit mode. If you give one, it is only a hint for the reserve estimate. The output keeps the structure of the source, so a source clip at 6 seconds returns about 6 seconds.
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: video-router-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"
}'Rates for the edit
Omni rates per second on Sume are $0.0375 at 360p, $0.125 at 720p, $0.1875 at 1080p and $0.375 at 4K. An edit at the default 720p of a 6-second source is 6 x $0.125 = $0.75. Native audio is always on, so the edit keeps a track. If the edit result is wrong, an exact retry with the same Idempotency-Key returns the original job, so change the key when you change the prompt.
Source clip requirements
The source should be a public HTTPS file you control. Keep it short: because the output follows the structure of the source, a long source is a long, expensive job. If you only need to change the last three seconds, trim the clip first with the video trim API at $0.02 per job and edit the trimmed file.
Do not expect the edit to change framing. Because aspect_ratio is rejected, the ratio comes from the source, so an edit of a 16:9 clip stays 16:9.
Sources
Related posts
More in Developers
- Omni reference-to-video with 3 clips of 3 s: a 10 s output is $1.25
Gemini Omni Flash 1.1 takes up to 10 reference images and 3 reference clips of 3 seconds each. A 10-second 720p output is 10 x $0.125 = $1.25 on Sume.
- generation_limits missing from a Sume submit: send one, then re-read
The docs say the snapshot is included when Sume can compute it. A Python rule for waves when it is absent, full, or open, tested on three inputs.
- HappyHorse 1.1 [Image 1] tags vs Sume's <IMAGE_REF_0> in prompts
HappyHorse reference-to-video counts images from [Image 1]; Sume's Omni counts from <IMAGE_REF_0>. How to map a prompt without an off-by-one.
- 8 active and 15 queued on Startup: the new in-flight budget is zero
Sume's headroom is min(max(0, concurrency - active - queued), queue_capacity_remaining). A Startup example that returns 0, the docs example that returns 60.
Written by Sume