Gemini Omni Flash
Released May 20, 2026
The first model in Google's Omni family: a multimodal model that generates video from mixed text, image, audio and video inputs and edits it conversationally over multiple turns.
- Price per second
- $0.10
- Max clip length
- 10 s
- Max resolution
- 720p
- Native audio
- Yes
- AA arena Elo
- —
- LMArena Elo
- 1513
- Capabilities
- text to video, video editing
- Licence
- Proprietary
Strengths & weaknesses
Strengths
- Conversational multi-turn video editing (element replacement, perspective changes)
- Combines text, image, video and voice references in one prompt
- Native sound generation: speech, music and sound effects
- Free for users 18+ in YouTube Shorts Remix and the YouTube Create app
Weaknesses
- 10-second generations
- No scene extension (added in Omni Flash 1.1)
- Gemini API endpoint shut down on 2026-10-22 (replaced by Omni Flash 1.1)
releaseDate is the consumer launch (Gemini app, Flow, YouTube) announced in Google's I/O 2026 roundup dated May 20 2026; the 'Introducing Gemini Omni' post itself carries no date. API public preview followed on June 30 2026 (gemini-omni-flash-preview; Vertex retirement date June 30 2027). Image-to-video conflict: the developer blog says it takes text, image and video inputs, but the Vertex model page marks 'Image to video' as not supported, so image inputs are treated as references and image-to-video is left out. maxResolution 720p: Google Cloud's text-to-video docs say 'gemini-omni-flash-preview only supports 720p', and the Gemini API changelog says the preview launched with 3–10 second videos at 720p. Not on Artificial Analysis' board. LMArena name 'gemini-omni-flash' (#2, Sep 21 2026). The Gemini API id gemini-omni-flash-preview is shut down: the deprecations page lists shutdown 2026-10-22 with replacement gemini-omni-1.1-flash. The Vertex retirement date (June 30, 2027) is unchanged as far as found.
Links
API model strings
- google-gemini-api
gemini-omni-flash-preview - google-vertex
gemini-omni-flash-preview