by Google
Any-input-to-video with voice and character consistency.
// Real output from Gemini Omni — generated through the API
The Gemini Omni family ships in tiers. Prototype on the cheapest, lock your prompt, then switch to the top tier for production.
cheap to prototype → premium to ship
Prototype on the cheapest tier, then switch the model_id to the top tier for production — same call.
Call the Gemini Omni API through your SamAutomation key — submit a job with the model_id, then poll until it's done. One key for every model, no per-vendor onboarding or separate billing.
Yes. On SamAutomation Gemini Omni Video runs through one endpoint - POST /api/ai/jobs with model_id "gemini-omni-video" - so you do not need a separate account or key with Google. You submit a job and poll it until the video is ready.
Gemini Omni Video starts at 90 credits per job at its cheapest settings. Longer clips, higher resolution and extra options cost more. Credits come with a plan, and the same key also covers JSON-to-video rendering and every other model.
No. Generation burns real GPU time, so every render costs credits - 90 at the floor for Gemini Omni Video. Anything advertising unlimited free generation is either putting you behind a long queue or watermarking the result. Plans bundle monthly credits instead.
Start at the cheapest tier in the ladder above and move up only when you can see the difference in your own output. Higher tiers cost more credits per job, and on short social clips that difference is often invisible.
You reach Gemini Omni Video through this API with your SamAutomation key, not through Google's own app, so that app's sign-up and regional availability are not what gates you. The model's own content rules still apply to what you generate.
Google publishes the model itself; this page is about calling it as an API alongside every other model on one key, which is a different job from their consumer app.
There is nothing to download. The Gemini Omni Video API documentation you need here is our own reference: one endpoint, one key, and the model_id above. Google publishes its own docs for the model, but those describe their platform, not this one.
If it is listed on this page it is active on the platform and you can call it today. Model availability moves — makers retire versions and add tiers — so the list here is the current state rather than a promise about next quarter.
You do not need a vendor SDK here. Gemini Omni Video is an HTTP call: POST /api/ai/jobs with your key and model_id "gemini-omni-video", then poll the job. Any language that can send JSON can drive it, and the request shape is the same for every model on this platform.
The practical questions are the same for every generation service: what the provider may do with what you upload, how long it is kept, and whether it can be used for training. Read that clause before you send client material through any model, ours included. Google publishes its own terms for the model itself.
That is up to Google, not to us, and it differs per model and per release. If open weights matter to you, check the maker's own licence. Running it here is renting inference, not getting the weights.
No. The consumer app and the API are separate products with separate limits. Here you call Gemini Omni Video through this platform, so there is no app to install and no second subscription.
Any-input-to-video with voice and character consistency. Pick it when that is the job; a cheaper model wins when it is not.
One key for JSON-to-video and every AI model — Gemini Omni Video starts at 90 credits. Documented and ready for your pipeline.