Google rolls out Gemini Omni 1.2 Flash with 15-second AI video generation

Google has pushed out Gemini Omni 1.2 Flash, the newest version of its multimodal video model. It is now generally available across the Gemini API and Gemini for business.

The headline upgrade is length. Native video generation now runs to 15 seconds, up from 10 seconds in earlier versions.

What changed in 1.2 Flash

Google announced the update on November 5, 2026. Three improvements lead the release: native 15-second video creation, multi-turn conversational editing, and stronger adherence to reference material.

Multi-turn conversational editing means a user can refine a clip through back-and-forth prompts instead of starting over each time. The model draws on context from earlier exchanges to keep scenes consistent across edits.

Reference following also gets a boost. Users can now feed in 7 to 10 reference images within a single generation process.

Output quality climbs as well. The model supports video generation up to 4K resolution, and clips come with synchronized audio rather than silent footage that needs a separate soundtrack.

Every generated video carries Google’s built-in SynthID watermark.

Pricing and where to get it

Access comes through several doors. Developers can use the Gemini API and Google AI Studio, while creators can reach it through Google Flow. Third-party platforms Runware and Segmind also offer the model.

On the consumer side, some access is free for eligible users in environments such as YouTube Shorts. Subscription tiers, including Google AI Plus, Pro, and Ultra, provide another route.

API pricing for 720p video output reportedly ranges from $0.02 to $0.10 per second. At that rate, a full 15-second clip at 720p would land somewhere in the range of $0.30 to $1.50.

How the Omni family got here

The Gemini Omni line debuted on May 19, 2026, at Google I/O. The 1.1 Flash release added scene extension and first-and-last frame control. The second feature lets users pin down how a clip begins and ends, with the model filling in the motion between. Version 1.2 builds on those tools and adds greater control over output resolution.

What this means for creators and competitors

For creators, the practical gains stack up. Longer native clips mean fewer stitched-together segments. Conversational editing means less re-prompting from scratch. More reference images mean tighter visual control over brand assets and recurring characters.

The general availability label also matters for businesses. Making the model generally available through the Gemini API and Gemini for business signals that Google considers it ready for production workloads, not just experimentation.

The third-party availability through Runware and Segmind suggests Google wants the model embedded in other developers’ pipelines, not just its own apps.

API pricing for 720p output ranges from $0.02 to $0.10 per second, and teams generating video at scale will need to model expenses carefully as clips get longer and resolutions climb toward 4K.