Google Releases Gemini Omni 1.2 Flash in Flow With 15-Second Shots and Per-Second Pricing

Google rolled out Gemini Omni 1.2 Flash in Google Flow, extending continuous shots to 15 seconds for Google AI subscribers and introducing per-second pricing for Omni 1.2 Flash generations, according to the Flow team's post. Longer takes let more of a scene play out in one generation, and you pay for the seconds you generate, edit, extend or upscale.

• Shot length depends on your plan. Every user can generate up to 10 seconds per shot, while Google AI Plus, Pro and Ultra subscribers can go from 11 to 15 seconds.
• You pay for each second of output. At 720p, a shot costs 3 Flow credits per second up to 10 seconds and 6 credits for each extra second from 11 to 15.
• Extend accepts reference images. Subscribers can lengthen a clip and steer what happens next with up to three images.
• Google claims cleaner output. The Flow team says all users get stronger visual consistency, fewer jump cuts, better reference adherence and more natural speech.
• The API price drops 30%. Developers pay $0.07 per second of 720p video, 30% less than Gemini Omni 1.1, per Google's model announcement.

360p drafts cost a third of 720p, and seconds 11 to 15 bill at double rate

A 10-second Omni 1.2 Flash shot costs 10 credits at 360p or 30 credits at 720p, according to Flow's credit table. Pushing that 720p shot to 15 seconds doubles the cost to 60 credits, because seconds 11 through 15 are billed at twice the base rate.

Google positions 360p as a draft tier at one-third the credit cost of 720p. Subscribers can then upscale Omni 360p videos to 1080p, or to 4K on Ultra, and extensions of 3 to 10 seconds cost 1 credit per added second at 360p and 3 credits at 720p.

Everyone using Google Flow gets 50 credits per day, which do not roll over. Monthly subscriber allotments on top of that:

• Google AI Plus: 200 extra credits
• Google AI Pro: 1,000 extra credits
• Google AI Ultra $100: 10,000 extra credits
• Google AI Ultra $200: 25,000 extra credits

Non-subscribers may be unable to generate video during peak hours, roughly 2 PM to 5 PM UTC, though they can still generate images and use the Flow Agent, which arrived in Flow alongside a way to build custom creative tools.

Omni Extend replaces Veo Extend as older Flow video models head for retirement

Omni Extend adds 3 to 10 seconds to a clip and accepts up to three reference images to guide the continuation. It requires a Google AI Plus, Pro or Ultra subscription and is available only in Flow on the web.

Google has retired Veo Extend, according to Flow's model guide, so extending a video in Flow means using Omni 1.2 Flash on a paid plan.

Veo 3.1 Fast, Veo 3.1 Quality and Omni 1.1 Flash remain usable until their December 5, 2026 retirement deadline. Google recommends switching to Omni 1.2 Flash or Veo 3.1 Lite for future projects.

Veo 3.1 Lite stays in Flow as a low-cost 720p option with 4, 6 and 8 second lengths, but it does not support video-to-video editing or extensions. Omni 1.2 Flash handles text, frames, ingredients and video-to-video editing in both aspect ratios.

The same model ships to the Gemini API, Google Vids and platforms like Runway

Outside Flow, Gemini Omni 1.2 Flash is generally available through the Gemini API, Gemini for business and Google Vids. Google Cloud's launch post adds multi-turn conversational editing, such as swapping a product, changing lighting or changing camera angle on the same clip, plus drafts at 360p or 720p that upscale to 1080p or 4K.

Outputs from the Gemini Omni model carry an imperceptible SynthID watermark, and audio references are gated behind allowlisting, according to Google Cloud.

Google Cloud names Artlist, FLORA, invideo, Krea, Picsart and Runway's AI tools among the platforms using the model. Shivam Rastogi, VP of Engineering at invideo, told Google Cloud that "15 seconds means a full scene in one go."

We ran Google Omni against Runway Aleph 2 on the same shots in a side-by-side test.