Consumer creation
Gemini app
Google lists Gemini as a place to try Omni. A Google AI subscription may be required, and features vary by tier and geography.
Google video model · checked July 30, 2026 Developer public preview
Quick answer
Gemini Omni Flash turns mixed inputs into editable video.
The public-preview model accepts text, image, and video context for ten-second video generation and natural-language edits. Treat product access and API integration as separate decisions.
Verified provider record
Provider status
Public preview for developers as of June 30, 2026. Preview lifecycle and limits can change.
API model ID
gemini-omni-flash-preview. Verify the live Gemini API documentation before shipping or storing the identifier.
Inputs and output
Combines text, image, and video inputs for video generation and conversational editing.
Published API price
$0.10 per second of video output at launch, described by Google as the same as Veo 3.1 Fast. Verify live pricing.
Current output length
Google documents 10-second video generations at launch, with longer duration described as future work rather than a current guarantee.
Flowith access
Not asserted here. Google product and API availability do not establish a Flowith model mapping.
Access map
One model family, different product contracts
Creative studio
Google Flow
Use the product interface for its supported creative workflow; do not infer API identifiers, billing, or limits from the UI.
Developer playground
Google AI Studio
Test the preview model and prompt pattern before integration. Playground availability is not a production SLA.
Programmatic preview
Gemini API
Use the documented preview model ID and current regional limitations. Re-check before production deployment.
Enterprise builder surface
Enterprise Agent Platform
Enterprise product availability and governance belong to that Google surface, not to the consumer app or Flowith.
Preview limitations to design around
- Video output is currently limited to 10 seconds.
- Audio-reference upload is not supported in the Gemini API for this model at launch.
- Scene extension is not supported in the Gemini API at launch.
- Video references up to three seconds can be accepted by the API schema but Google says the model does not process them correctly at launch.
- Character consistency can weaken across scene changes or camera pans.
Official sources
- Google developer launch — API model ID, price, inputs, output length, limits, and preview access.
- Google DeepMind model page — product surfaces, model positioning, benchmarks, and subscription boundary.
- Gemini Omni API documentation — verify current regions, requests, limits, model lifecycle, and prompting.
Continue by decision intent
Product vs API access Separate Gemini, Flow, AI Studio, API preview, subscriptions, geography, and Flowith status. API preview workflow Plan inputs, ten-second output, iterative edits, review gates, and preview migration. Veo 3.1 Compare a specialized Google video route without treating the launch price as capability parity. Gemini 3.6 Flash Switch to a workhorse assistant model when video output is not the primary task.
Gemini Omni Flash questions, answered
Gemini Omni Flash is Google's natively multimodal video model for generation and conversational editing from text, image, and video inputs. Google opened it to developers in public preview on June 30, 2026.
Google's launch documentation names gemini-omni-flash-preview. Because it is a preview identifier, verify the live Gemini API documentation before implementation.
Google published a launch price of $0.10 per second of video output. Check the current API pricing, taxes, failed-request treatment, storage, retries, and downstream processing before budgeting.
Google documents 10-second generations at launch. Longer durations were described as coming later, so do not treat them as currently available.
Google lists the Gemini app, Google Flow, Google AI Studio, Gemini API, and Gemini Enterprise Agent Platform, with other product surfaces shown on its model page. Availability can depend on subscription tier and geography.
This page does not claim a Flowith model mapping. Verify the current Flowith workspace separately from Google's consumer products and public-preview developer surfaces.