gemini-omni-1.1-flash vs gemini-omni-flash-1.1: the Sume id map
Google names it gemini-omni-1.1-flash; Sume's id is gemini-omni-flash-1.1. A small Python map from Google ids, Veo 3.1 previews included, that fails loudly.

Short answer
Google's Omni model id is gemini-omni-1.1-flash (Gemini Omni docs, read 2026-10-04). Sume's catalog id for the same family is gemini-omni-flash-1.1. The words are the same; the order is not, and Sume returns 404 model_not_found for any id that is not in the catalog. Copying the Google id into a Sume request will fail, so keep a tiny explicit map.
The map
Google's Veo 3.1 preview ids have a shutdown date of Oct 22, 2026, and its deprecations page recommends Omni as the replacement. Sume does not list Veo, so the three Veo ids map to Omni by recommendation, not by identical behavior.
| Google id | Status at Google | Sume id |
|---|---|---|
| gemini-omni-1.1-flash | Current | gemini-omni-flash-1.1 |
| veo-3.1-generate-preview | Shutdown Oct 22, 2026 | gemini-omni-flash-1.1 |
| veo-3.1-fast-generate-preview | Shutdown Oct 22, 2026 | gemini-omni-flash-1.1 |
| veo-3.1-lite-generate-preview | Shutdown Oct 22, 2026 | gemini-omni-flash-1.1 |
A map that refuses to guess
Do not fall back to a default silently. An unknown id should raise, so a typo is caught in tests and not in a bill.
import asyncio
SUME_ID = {
"gemini-omni-1.1-flash": "gemini-omni-flash-1.1",
"veo-3.1-generate-preview": "gemini-omni-flash-1.1",
"veo-3.1-fast-generate-preview": "gemini-omni-flash-1.1",
"veo-3.1-lite-generate-preview": "gemini-omni-flash-1.1",
}
def to_sume(google_id):
try:
return SUME_ID[google_id]
except KeyError:
raise ValueError("no Sume id for " + google_id)
async def main():
print(to_sume("veo-3.1-fast-generate-preview"))
asyncio.run(main())What changes besides the id
Sume's Omni accepts 3 to 10 seconds at 360p, 720p, 1080p and 4K in 16:9 or 9:16, with audio. Veo clips are 8 seconds, so an old Veo call with an 8 second duration is valid on Omni. Veo's extension flow, which adds up to 7 seconds at a time, has no direct equivalent in a one-shot Sume request; chain clips with frame_images or use the video edit field instead.
Pricing also changes: Sume bills provider list times 1.25, and Omni's list is $0.03 a second at 360p, $0.10 at 720p, $0.15 at 1080p and $0.30 at 4K, so Sume bills $0.0375, $0.125, $0.1875 and $0.375.
Where to keep the map
Put the map in one module and import it everywhere, so a rename is a one-line change. Add a unit test that fetches the Sume catalog in CI and asserts every value in the map is listed; if Sume ever changes an id, the build fails before production does. Log the Sume id you actually sent with each job id, so a cost or quality question months later has a clear answer.
Caveats
- Check
GET /v1/videos/modelsat startup and compare the ids in your map against it. - Ids can be renamed again; keep the map in one file.
- A model recommendation is not a quality guarantee; test your prompts on Omni before the Oct 22 date.
Sources
Related posts
More in Developers
- Gemini Omni API first-day checklist: billing, $10 window, $250 cap
Before your first Omni call on the Gemini API: link billing for Tier 1, know the $10 window and $250 cap, use the right model id, and test a 3-second 360p clip.
- Omni Flash prompts in English, captions in your language
Google says Omni fully supports English; other languages are unevaluated. Prompt in English, then add captions in your language with Sume's captions API.
- Gemini Omni Flash resolution: "4K" or "4k" on Sume
Sume accepts 4k as an alias of 4K for Gemini Omni Flash 1.1 and translates it to the lowercase token fal expects. Billed rate and Google's 4K caveat.
- Moving Veo scripts to Gemini Omni: an Interactions API checklist
Veo 3.1 previews shut down Oct 22. Omni uses the Interactions API, task values, base64 or uri delivery and different limits. A checklist for rewriting scripts.
Written by Sume