Omni replaces Veo in the Gemini app: does your API code change?

No: nothing in Google's announcement changes an API call by itself, because the sentence that matters is scoped to a product. Google's Gemini Omni page says Omni "will replace Veo in the Gemini app", which describes what the app generates with, not which model ids an integration may send. If your code calls a video model through an API, the thing to check is the catalog of the API you call.

On Sume the answer is short. The published video catalog lists no Veo id, and it lists one Omni row, gemini-omni-flash-1.1. A pipeline that sends Sume that id keeps working the same way before and after Google's app change.

What Google's page says, and what it leaves out

The page is a consumer product page, so it describes the app. It also does not say what happens to existing Veo projects, and it gives no resolution. Treat anything beyond the lines below as unconfirmed.

The Omni row you can call on Sume

Sume's video generation docs describe gemini-omni-flash-1.1 as 3 to 10 seconds at 360p, 720p, 1080p or 4K, in 16:9 or 9:16, with native synced audio that is always on. It takes text, a start frame and optional end frame, references (up to 10 images and up to 3 video clips of 3 seconds each), and a video_url for editing. It takes no audio references. Sume bills the provider list price times 1.25 per output second.

A short audit for your own code

• Search the repo for the model id string you send, not for the word Veo; the id is what a vendor retires.
• Read the live catalog with GET /v1/videos/models before you hard-code a duration or resolution, since limits differ per model.
• If a feature you need is only described on the app page, such as scene extensions, check the API docs for a field that does it before you plan around it. Sume's Omni docs list no extend input.
• Keep the id in config, so a swap is a one-line change and a re-run of your sample prompts.

Why the app and the API can diverge

An app picks a backend for you and can swap it overnight. An API gives you a named id, a documented envelope and a price per second. When a vendor says a model replaces another in its app, the useful question is whether a field you rely on, such as duration, resolution or a reference type, still exists in the API you call. For Omni on Sume those limits are the ones in the table above, and they are the same whatever the Gemini app does.

If your brief says Veo

A client brief may name Veo because that is the name they saw in the app. Rather than arguing about names, translate the brief into the properties it needs: clip length, aspect ratio, resolution, whether sound is included, and whether you must edit an existing clip. Then pick a catalog row that meets those properties. On Sume, an Omni row covers 3 to 10 second clips with sound in 16:9 or 9:16, plus an edit mode. If the brief needs something outside that, such as 15 seconds, the row table in Sume's docs shows which other ids reach it.