Migrate Veo 3.1 API calls to Gemini Omni Flash 1.1: parameter map
Veo 3.1 previews shut down October 22, 2026. A parameter-by-parameter map from the Veo guide to Gemini Omni Flash 1.1 requests on Sume, with a working curl.

To move a Veo 3.1 integration to Gemini Omni Flash 1.1 on Sume, change the model id to gemini-omni-flash-1.1, send durations of 3 to 10 whole seconds instead of 4, 6 or 8, move reference media into input_references, and drop any audio-off setting, because Omni's audio is always on. The Veo preview ids shut down on October 22, 2026 according to Google's deprecations page, read 2026-10-01.
This map compares Google's Veo 3.1 guide (read 2026-10-01) with Sume's Video Generation and Video Router docs.
Which Veo setting becomes which Sume field?
Sume's POST /v1/videos follows the OpenRouter video wire, so the field names are duration, resolution, aspect_ratio, frame_images and input_references. Mode is inferred: frame images mean image-to-video, references mean reference-to-video, and if both are present frame_images wins.
| Topic | Veo 3.1 guide | Omni Flash 1.1 on Sume |
|---|---|---|
| Duration | 4, 6 or 8 seconds; 8 required for 1080p, 4K and reference images | 3 to 10 whole seconds |
| Resolution | 720p, 1080p, 4K (1080p and 4K at 8 seconds only) | 360p, 720p, 1080p, 4K |
| Aspect ratio | 16:9 or 9:16 | 16:9 or 9:16 |
| Reference media | Up to three images | Up to 10 images and up to 3 clips of 3 seconds or less |
| Audio | Generated natively | Native, always on; generate_audio: false has no effect |
| Seed | Not part of this map | Rejected: no Sume video model accepts seed |
What does a migrated request look like?
This is an image-to-video request with the first frame pinned. Idempotency-Key makes a retried submit return the original job instead of creating a second paid one. Poll the returned polling_url or pass an HTTPS callback_url.
curl -X POST https://api.sume.com/v1/videos \
-H "Authorization: Bearer $SUME_API_KEY" \
-H "Content-Type: application/json" \
-H "Idempotency-Key: veo-migration-001" \
-d '{
"model": "gemini-omni-flash-1.1",
"prompt": "Slow push-in on the product, soft window light",
"duration": 8,
"resolution": "1080p",
"aspect_ratio": "9:16",
"frame_images": [{
"type": "image_url",
"image_url": {"url": "https://example.com/first.png"},
"frame_type": "first_frame"
}]
}'What is not a one-to-one move?
Reference clips are shorter on Omni (3 seconds each) and are addressed in the prompt as <VIDEO_REF_0>, which has no Veo equivalent. Veo's extension feature, 7 seconds per step, has no matching Sume input in the docs. Veo's two-day video retention does not apply: a completed Sume job gives you download URLs on the job, and the Sume docs tell integrations to store the Sume URL.
Run a handful of real prompts on both sides before cutover. Nothing here says Omni will reproduce a Veo look, only that the request shape maps.
A cutover checklist
Do these in order so a surprise shows up in staging rather than in a customer's queue.
- Swap the id to
gemini-omni-flash-1.1and readsupported_durationsandsupported_resolutionsfromGET /v1/videos/modelsbefore you hard-code anything. - Replace any 4, 6 or 8 second constants with a value between 3 and 10, and drop
generate_audio: false, which does nothing here. - Move reference images into
input_referencesand trim reference clips to 3 seconds or less. - Send an
Idempotency-Keyon every submit, and read the final amount fromusage.coston the completed job. - Keep the old Veo path behind a flag until the date passes, then delete it.
Sources
Related posts
More in Developers
- MiniMax H3-Context-IR returns a prompt, not a video: Sume view
MiniMax's H3-Context-IR endpoint reads text, images, audio and video and returns an enhanced prompt without making a video. What Sume has instead.
- Fine-tune MiniMax H3 with a LoRA: data grid, recipe and serving
MiniMax points to an experimental miles-diffusion LoRA recipe for H3: 1344x768, 24 fps, 17n+5 frames, 8 x H200. The data rules and how to serve it.
- MiniMax H3 on SGLang: 8 x B200, FL2VA server and /v1/videos
MiniMax documents a preview SGLang server for H3: 8 x B200, Ulysses 8, about 19 s per clip. The setup, the memory numbers, and when to host instead.
- MiniMax image-01 API: 9 images, 1,500 characters, 24-hour URLs vs Sume
MiniMax image-01 takes a 1,500-character prompt, 1 to 9 images, 512-2048 px sizes and returns URLs that expire in 24 hours. Sume's Image API limits compared.
Written by Sume