Google released Gemini Omni 1.2 Flash on November 5, 2026, extending native video generation to 15 seconds and reducing 720p pricing by 30 percent.
The updated multimodal model from Google DeepMind is now generally available via the Gemini API and Google Cloud. It integrates into Google Vids, Flow, Google Ads, and the Gemini app for Google AI Ultra subscribers, according to Google.
Google Reduces Video Clip Lengths and Pricing
The new version increases the maximum length of natively generated and edited clips from 10 seconds to 15 seconds.
Pricing for 720p video has dropped to $0.07 per second. This represents a 30 percent decrease compared to Gemini Omni 1.1. A full 15-second clip at 720p resolution costs approximately $1.05, MarkTechPost noted.
Google suggests a draft-then-upscale
workflow to manage expenses. Developers can create initial drafts at lower resolutions like 360p or 720p. They can then use follow-up edits to correct specific details. Finally, they can upscale the preferred result to 1080p or 4K, The Decoder reported.
New Tools Improve Visual Consistency
A central feature of the update is Conversational Editing.
This allows users to modify elements of a generated clip through a dialogue. Users can change the background, lighting, or aspect ratio without regenerating the entire video, The Decoder reported. The documentation shows this is achieved through a previous_interaction_id,
which prevents the need to re-upload the prior video.
The model also aims to improve reference adherence. It combines character or product images with audio cues and motion clips of up to three seconds. The AI is designed to keep faces, clothing, packaging typography, and brand colors consistent across different camera angles, Google stated. This facilitates tasks like creating virtual try-on clips, catalog spins, and branded video ads.
Google provided an example involving a 15-second sequence of a model performing 360-degree turns. This sequence preserves the model’s identity and the specific cut and drape of three different outfits.
Omni 1.2 Flash produces more natural-sounding dialogue with clearer lip-sync, Google reports. The model can also distinguish between multiple speakers. It handles continuous camera movements like dramatic zoom-outs or FPV drone flights. These movements show a reduction in unwanted jump cuts. The model is also expected to show less visual drift or morphing during transitions than the previous version.
Regarding the difficulty of professional video production, Omar Pera, CPO at Magnific, said:
The first generation is a starting point. Professional teams need precision to get from there to a finished shot.
Omar Pera, CPO at Magnific
Model Performance Shows Positive Feedback and Technical Limitations
No specific benchmark scores for version 1.2 have been published. However, the DeepMind model page reports human-rater results for the Omni Flash line using 1,003 MovieGenBench prompts. An earlier Omni Flash version reached the number one position on the Arena text-to-video board in June.
Exciting news: Gemini Omni Flash is now #1 in the Video Arena (both Text-to-Video and Image-to-Video)! For Text-to-Video this is a massive +158 pt improvement over Veo 3.1 (1080p) and a large +61 pt lead over the next best model, Seedance 2.0. Congrats @GoogleDeepMind for this… https://t.co/xPQSUkB0lG pic.twitter.com/MAo6JufDbj
Early feedback from partners has been positive, though it is noted as being self-selected. Google has acknowledged specific technical limitations still under development. These include the management of complex physics, rapid changes to scripts, and fast sports action. Users of the Gemini API should also note that audio references are currently allowlist-gated on Google Cloud.
Anyone using Gemini Omni 1.1 via the API should prepare for a change. Starting December 3, 2026, gemini-omni-1.1-flash will be redirected to the new model via a fixed pipeline.
Users Access the Model Through Various Platforms
Integration varies depending on the platform. In Google Vids, eligible users with a Google account or Workspace subscription can generate up to three 15-second videos each month at no additional cost, NeoTeo reported. This monthly allowance is for Google Vids accounts and is not a general allowance for all Google video-generation products.
For developers, the model is accessible via the Gemini API and Google AI Studio. Businesses can utilize the model through the Google Cloud Console using Provisioned Throughput for predictable capacity, Google confirmed. Gemini Omni 1.2 Flash introduces longer outputs and better adherence to reference images.
All generated outputs include an invisible SynthID watermark to verify their AI origin, MarkTechPost reported.