Gemini Omni Flash Public Preview: What Changed for Developers
gemini-omni-flash-preview. This page has been updated from its earlier pre-launch watch status.On EvoLink, teams can use four task-oriented route IDs for text-to-video, image-to-video, reference-to-video, and video editing while keeping the same async task workflow used by other video models.
Confirmed API status
| Item | Current verified status |
|---|---|
| Google lifecycle | Public preview |
| Official model ID | gemini-omni-flash-preview |
| Output | 3–10 second, 720p, 24 FPS video |
| Aspect ratios | 16:9 and 9:16 |
| Tasks | Text-to-video, image-to-video, reference-to-video, edit |
| Official video price | $17.50 per 1M video tokens, about $0.1014 per second |
| Free API tier | Not available for this model |
Official model ID vs EvoLink route IDs
gemini-omni-flash-text-to-video, gemini-omni-flash-image-to-video, gemini-omni-flash-reference-to-video, and gemini-omni-flash-video-edit. These are EvoLink route IDs, not alternate Google model IDs.Limits to design around
The current API does not support uploaded audio references, multi-video reasoning, video extension, interpolation, or voice editing. Video references up to three seconds may be accepted by the schema but are not processed correctly. System instructions, temperature, top-p, stop sequences, and a separate negative-prompt field are unsupported. Put exclusions directly in the prompt. Uploaded-video editing is also restricted in the EEA, Switzerland, and the United Kingdom. Every generated video includes invisible SynthID provenance.
Because the model is still in preview, production teams should keep retries, cost logging, moderation handling, and a fallback route. EvoLink's unified API lets the application compare or switch to Veo, Seedance, Kling, and other routes without rebuilding the user workflow.
Try Gemini Omni Flash on EvoLinkFAQ
Is Gemini Omni Flash API available now?
Yes. It is available in public preview through the Gemini API.
What is the official model ID?
gemini-omni-flash-preview.How much does it cost?
Google's effective 720p video-output rate is about $0.1014 per second; input and text or reasoning output are additional.
What output does it support?
It produces 3–10 second, 720p, 24 FPS videos in 16:9 or 9:16.
Can it edit video conversationally?
previous_interaction_id.Can I upload an audio reference?
No. The model can generate audio with a video, but uploaded audio references are not supported in the current API.
Does it support 1080p?
Google's current Gemini API model page confirms 720p, not 1080p.
Is it production-ready?
It is a preview model. Test quality, latency, safety blocks, retries, and fallback behavior before production rollout.
