Google Introduces Gemini Omni 1.2 Flash: 15-Second AI Videos and a 30 Percent Price Cut

In the world of generative AI, text-to-video technologies are quickly evolving from short-lived, high-cost experiments into full-fledged production tools. Taking one of the strongest steps in this transformation, Google has officially released Gemini Omni 1.2 Flash, its native multimodal video generation and editing model. Built on top of the previous-generation Omni 1.1 architecture, the new model increases the video length that can be generated in one go by 50 percent to 15 seconds, while cutting API usage costs for developers and enterprises by 30 percent.

Optimized for content creators, advertising agencies and software developers who want to build high-quality visual storytelling in a short time, Gemini Omni 1.2 Flash does more than just produce longer clips. With multi-turn, chat-based editing capabilities that remove the need to re-render from scratch and advanced visual consistency mechanisms, it also resolves the biggest bottlenecks in commercial video workflows.

From 10 Seconds to 15 Seconds: Flexible Duration Control and Fluid Storytelling

One of the biggest limitations of AI-powered video generation models was that the clips they created were only short snippets of a few seconds. The native single-pass video generation and editing capacity, capped at a maximum of 10 seconds in the previous version Gemini Omni 1.1 Flash, has been raised to 15 seconds with Gemini Omni 1.2 Flash. On top of that, users and developers can set the length of the videos they generate precisely to their needs, between 3 and 15 seconds in 1-second increments.

At first glance a 5-second increase may look like a small technical detail, but it represents a critical threshold for cinematic storytelling and commercial production:

• Social Media and Digital Ad Formats: 15-second spots, the basic building block of standard ad and promo edits on platforms such as Instagram Reels, YouTube Shorts and TikTok, can now be produced seamlessly in a single generation step.
• Timing and Camera Movement Integrity: Scene transition breaks, physical motion inconsistencies and lighting glitches that appear when trying to combine several short clips in external editing software disappear entirely thanks to 15-second single-piece generation.
• Precise Budget and Duration Management: Choosing the duration in 1-second steps lets developers avoid unnecessary compute costs by generating video only as long as the script requires.

A 30 Percent Price Cut and a Smart Draft-to-4K Workflow

Without a doubt, one of the biggest obstacles to adopting generative video models at enterprise scale was high GPU and API costs. With the Gemini Omni 1.2 Flash release, Google made a rather aggressive cut to its pricing policy. The per-second cost of standard 720p video generation, offered through the Gemini API and Google Cloud, was cut by 30 percent, from about $0.10 to $0.07. With this new rate, the total generation cost of a full-length 15-second HD video clip stays at just $1.05.

Google also recommends a staged production workflow it calls “Draft-to-Upscale” for teams that want to use their production budgets even more efficiently. Under this approach, creative teams create their first ideas and scene trials very quickly and cheaply at lower native resolutions such as 360p or 720p. Once the scene composition, camera angle and motion dynamics are approved, only the selected final video clip is upscaled to 1080p or 4K resolution to reach broadcast quality. This prevents high-resolution render budget from being spent during the trial-and-error stages.

Editing Video by Chatting Instead of Regenerating from Scratch

In traditional text-to-video tools, when you want to change a small detail in a generated video, you have to update the prompt and regenerate the whole scene from scratch. This process both wastes time and carries the risk of losing character expressions or camera movements you liked in the previous attempt.

Thanks to its multimodal structure, Gemini Omni 1.2 Flash integrates the multi-turn conversational editing experience directly into video editing. Users can revise a video they have created step by step by giving natural language commands through the chat interface:

• Completely changing only the background location while keeping the main character and their movement in the scene,
• Swapping the product in an ad video for another model or color,
• Turning the scene's lighting atmosphere from daytime to sunset or to a neon-lit night setting,
• Repositioning the camera angle, framing or lens focal length without disrupting the overall flow of the scene.

This dialogue-based approach turns the AI model from a one-shot video generator into an interactive post-production assistant that responds instantly to the director's commands.

Multimodal Reference Consistency and Enterprise Ecosystem Integration

For commercial brands and professional studios, the most sensitive issue in AI videos is visual consistency. Text on a product's packaging getting garbled, brand colors drifting from scene to scene or a character's facial features changing as the angle changes makes professional use of the generated content impossible.

Gemini Omni 1.2 Flash is equipped with advanced reference adherence capabilities that can process text prompts, reference images, short motion clips and audio cues at the same time. Even when the camera angle or setting changes, the model can keep human faces, clothing textures, corporate brand colors and typographic text on product packaging stable with high consistency. Thanks to motion synchronized with audio cues, audiovisual coherence reaches a much more natural level.

Broad Access Across the Google Ecosystem and SynthID Safety

The new model is generally available to developers through Google AI Studio and the Gemini API, and to enterprise customers on Google Cloud infrastructure. On the end-user and business side, it is being gradually integrated into Google Vids, Google Ads, Flow and directly into the Gemini app. The Google Ads and Google Vids integrations in particular will let marketing teams turn campaign visuals into high-converting 15-second video ads within minutes.

On the safety and transparency side, Google's standard policy continues without compromise. A digital SynthID watermark, imperceptible to the human eye but detectable with dedicated verification tools, is automatically embedded in every frame of every video generated or edited with Gemini Omni 1.2 Flash. Even if videos are later cropped, compressed or color graded, the information that the content was generated by AI is securely preserved. Bringing together longer duration, lower cost and precise scene control, Gemini Omni 1.2 Flash proves once again that AI-powered video production has moved out of the experimental stage and become an industrial standard.