Gemini Omni Flash 1.1 on Sume: one model id, four request shapes

On Sume, Gemini Omni Flash 1.1 is a single catalog id, gemini-omni-flash-1.1. You never select an endpoint: Sume routes the request by the fields you send. A prompt alone is text-to-video, an image is image-to-video, reference lists are reference-to-video, and a video_url is an edit. This is the Video Router wire, at POST /v1/video-router/generate.

The four shapes

Limits are the Sume docs' own, read 2026-10-05.

Rules that catch people

• Native synced audio is always on, and the API rejects generate_audio: false.
• There is no bitrate_mode and no reference_audio_urls.
• video_url is the edit source, not a reference: you cannot send it with image_url, end_image_url or reference_*_urls.

An edit request

This is the docs' example, with the bottle swapped for an apple:

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"
  }'

Price and the vendor page

Sume bills the provider list times 1.25 per output second, by resolution: $0.0375 at 360p, $0.125 at 720p, $0.1875 at 1080p and $0.375 at 4K. Google's launch post for Gemini Omni 1.1 Flash (read 2026-10-05) describes scene extension up to 40 seconds in 10-second steps, a 360p draft mode and video references up to 3 seconds. Sume's catalog entry takes 3 to 10 seconds per job and 3-second video references; it does not list a scene-extension field, so for longer pieces plan several jobs.

Choosing between the shapes

Use text-to-video when you have no assets. Use image-to-video when you have an exact first frame, and add end_image_url if you also have a last frame. Use reference-to-video when you want a subject or style to carry through but not as an exact frame: up to 10 images and up to 3 videos of 3 seconds or less, referred to in the prompt as <IMAGE_REF_0> and <VIDEO_REF_0>. Use the edit shape when you already have a clip and want to change part of it.

Do not mix the edit field with others. The API rejects video_url together with image_url, end_image_url or reference_*_urls. If you try, expect a 400 and no charge, since the request is not accepted.

Before you ship anything, read the live pages again: the catalog is public, the pricing page is public, and the docs describe the request fields. A blog post is a snapshot. The catalog, the plan grid and the error table are the things that change, so write your code to read them instead of copying numbers from a page, and re-check when a new model is added.

A good habit is a small log line per submit with the model, resolution, duration, estimated cost, job id and the Idempotency-Key you used. When a job misbehaves, those six fields answer most of the questions support will ask, and they let you compare your estimate with usage.cost and the usage ledger without re-running anything.

If you are new to the API, start with one clip, one model and the lowest resolution, read the full response once, and only then build a loop around it. Most surprises with video jobs come from fields that were defaulted, not from fields that were set.