Omni is not Veo
Many developers expected Google I/O 2026 to bring Veo 4. Instead, Google introduced Gemini Omni, a family of multimodal Gemini models that can also produce video. The difference matters when choosing a model:
- Veo 3.1 is a specialized video generator focused on visual fidelity: textures, lighting, camera.
- Gemini Omni brings Gemini's world knowledge to video. It is better at getting things right: period costumes, architecture, scientific processes, on-screen logic.
As of early October 2026, Google has not announced a model called Veo 4. Third-party pages describing Veo 4 specs or prices are speculation.
What Gemini Omni Flash can do
- Generate 3 to 10-second clips with native audio from text or images.
- Edit short uploaded videos.
- Interpolate between a first and a last frame, and extend existing scenes.
- Output from 360p up to upscaled 4K.
Pricing: tokens, not seconds
Gemini Omni Flash uses Gemini's token billing. At standard rates, input costs $1.50 per million tokens, text and thinking output $9 per million, and video output $17.50 per million. A second of 720p video is about 5,792 output tokens, which puts 720p output at roughly $0.10 per second, or about $1 for a 10-second clip. There is no free API tier, although you can try the model in Google AI Studio.
Token billing makes costs harder to forecast than per-video pricing, because resolution and length change the token count. On VideoGenAPI, Gemini Omni is included in monthly plans and billed per video, at up to 1080p for 4 to 10 seconds.
When to choose Gemini Omni over Veo
| Your brief | Pick |
|---|---|
| Historical, educational or scientific scenes that must be accurate | Gemini Omni |
| Premium commercial look, landscapes, interiors | Veo 3.1 Fast |
| Fast, affordable realistic clips with audio | Veo 3 Fast |
| People moving, choreography | Kling 3 (not Google) |
Calling Gemini Omni through VideoGenAPI
{
"model": "gemini-omni",
"prompt": "Ancient Roman forum at midday, citizens in togas debating, accurate architecture for 100 AD",
"duration": 8
}No Google Cloud project, Vertex AI setup or token accounting is needed. Prompts can be up to 2,000 characters for this model, double the default, which suits its knowledge-heavy use cases.
Frequently asked questions
When did Gemini Omni Flash become available?
Is Gemini Omni the same as Veo 4?
How much does Gemini Omni Flash cost?
How long are Gemini Omni Flash videos?
Sources
Third-party facts and prices were checked on the date of publication and can change; check the provider before buying.