Gemini Omni Flash: Google's Video Model, and a Preview Shutting Down

The preview endpoint is being switched off

Before anything else: if your code calls gemini-omni-flash-preview, it stops working on 30 September 2026. The generally available model, gemini-omni-1.1-flash, has been live since 27 August 2026, and the migration is a model-identifier change rather than a rewrite.

What Gemini Omni Flash is

Gemini Omni Flash is Google's fast video generation model, first released in preview on the Gemini API on 30 June 2026. Google describe it as covering video generation, editing, keyframe interpolation and extension, with native audio — meaning the audio is produced as part of the generation rather than added afterwards.

Pricing

Billing is per video output token, at $17.50 per million, which Google's own materials translate to roughly $0.10 per second of video. A one-minute clip is therefore around $6. There is no free tier.

LMArena

On LMArena's text-to-video leaderboard Gemini Omni Flash has 1513 Elo (rank 2). On the other video boards it has 1465 Elo in Image to Video and 1367 Elo in Video Editing.

Capabilities

Google list text-to-video generation, editing, keyframe interpolation and extension, with audio generated as part of the video. The preview identifier gemini-omni-flash-preview stops working on 30 September 2026.

The June 2026 preview

Google released Gemini Omni Flash in public preview on 30 June 2026 in Google AI Studio and the Gemini API, and in the Gemini app and Google Flow, describing it as a model for video generation and conversational editing. It accepted text, image and video input and generated 10-second videos at launch, priced at $0.10 per second of video output. Google listed limitations at preview: audio references and scene extension were not supported, video references of up to 3 seconds were accepted by the API but not correctly processed, and character consistency could slip across scene changes or panning. Output carries SynthID watermarking, which can be checked through the Gemini app, Gemini in Chrome or Search.

Gemini Omni 1.1 Flash

Google released Gemini Omni 1.1 Flash on 27 August 2026 as the generally available version, identifier gemini-omni-1.1-flash. Scene extension analyses up to 10 seconds of prior context and extends a video in 10-second increments up to 40 seconds in total. First and last frames can be specified for transitions between keyframes. Output resolutions are 360p, 720p, 1080p and 4K; Google state the 360p draft setting generates up to 60% faster than the standard 720p at a third of the cost. Video references of up to three seconds are supported as input. It is available in Google AI Studio, the Gemini Enterprise Agent Platform API, Google Flow for Google AI Plus, Pro and Ultra subscribers, and the Gemini app.

In Google Vids

On 23 September 2026 Google brought Gemini Omni 1.1 Flash to Google Vids, stating that anyone with a Google or Google Workspace account can generate videos there at no cost, with expanded capacity on paid Google AI plans and Workspace Business and Enterprise plans. Vids generates new scenes in 1080p, can upscale existing AI clips, and supports scene extension and exact duration control. Every clip generated with Omni 1.1 in Vids carries an imperceptible SynthID watermark embedded in the video frames.

Official Resources