Google Gemini Omni 1.2 Flash Brings 15-Second Clips at $0.07 a Second
Highlights
- Gemini Omni 1.2 Flash is generally available on the Gemini API and Gemini for business
- Clips now run up to 15 seconds, up from 10 seconds in the prior version
- 720p video costs $0.07 per second, 30% cheaper than the prior version
- Finished clips can be edited by chatting, such as swapping background or lighting
Google DeepMind said on 5 November that Gemini Omni 1.2 Flash, its video generation model, is generally available on the Gemini API and Gemini for business, according to the Google DeepMind announcement, 5 November 2026. The update focuses on the two things creators ask for most: longer clips and a lower price.
15 seconds, in one-second steps
Clips can now run up to 15 seconds instead of 10, in one-second steps from 3 to 15, according to the Google DeepMind announcement, 5 November 2026. Feed it reference images, short motion clips and audio cues, and Google says faces, clothing, packaging text and brand colours stay consistent across camera angles, the summary said. Fifteen seconds is long enough for a full advertisement spot in a single generation.
Price cut and chat editing
720p video costs $0.07 per second, 30% below the prior version, according to the Google DeepMind announcement — at that rate a full 15-second 720p clip costs $1.05, Daily AI Thread calculated. Users can also edit a finished clip by chatting, such as swapping the background or lighting, across the full 15 seconds without regenerating it, which Google pitches at teams making advertisement variants for different audiences. Google itself says fast sports action and complex physics are still being improved, Daily AI Thread noted.
The launch lands in a crowded week for AI video and agents, alongside Tencent’s Hy4 Can See and Use a Computer — but Its Open Weights Are Not Out Yet, and it underlines how quickly consumer-facing AI pricing is falling even as enterprise costs rise — the backdrop to Apple Cuts Engineers’ Daily AI Budget From $300 to $200 as Token Bills Bite.
For small studios and social teams, the practical test will be consistency: if faces and brand colours really hold across angles at $0.07 a second, short-form advertisement production gets considerably cheaper. Google’s own caveat on fast action and physics is the part to watch in early user tests.
What this means
The significance of this release is economic as much as technical. Fifteen-second clips, adjustable in one-second steps from three seconds up, cover the length of a typical social ad or product spot in a single generation, and at $0.07 per second for 720p — a 30% price cut on the previous version — a full clip costs $1.05. Numbers like that move AI video from experiment to line item: a marketing team can generate dozens of variants for the price of one traditional reshoot. The second shift is workflow. Multi-turn conversational editing — swapping a background, changing the lighting, switching aspect ratio in chat — removes the regenerate-from-scratch lottery that has made AI video expensive in practice even when the sticker price looked low.
Against alternatives, Google’s pitch is iteration speed and cost rather than spectacle. Its own workflow guidance says to draft fast at 360p or 720p, fix details with a follow-up edit, then upscale only the best creations to 1080p or 4K. The comparison with the previous Omni 1.1 is direct: ten seconds becomes fifteen, and reference adherence is claimed to hold faces, garments, packaging typography and brand colours consistent across camera angles — the failure mode that has kept brands away from AI product video. Google is candid about the remaining edge cases, saying fast sports action and complex physics are still being improved, which is where premium, spectacle-first video tools keep their advantage for now.
For US readers, availability is immediate and broad: generally available through the Gemini API and Gemini for business, usable in Flow, in Google Vids — where Google says 15-second clips come at no additional cost with a Google account or Workspace subscription — in the Gemini app for AI Ultra subscribers, and in Google Ads campaigns. Outputs carry an imperceptible SynthID watermark, which matters for any US business that has to disclose AI-generated advertising. What to watch is whether the consistency claims survive production use by real brands, because that — not the demo reels — is what decides whether the $0.07 second becomes the industry’s price to beat.
The real cost comparison for a US marketing team is not a single clip against a single shoot. It is the full cycle of producing enough usable variants for different audiences, checking each one for mistakes, and revising the versions that are almost right. That is why conversational editing matters alongside the headline price. If a team can keep a finished fifteen-second clip and change its background, lighting or aspect ratio in chat, it preserves the parts that already worked instead of paying again — in money, time and randomness — for a completely new generation. Google’s suggested workflow points in the same direction: draft quickly at a lower resolution, fix details through follow-up edits, and upscale only the strongest results. Used that way, the $0.07-a-second rate is the start of the saving, not the whole of it; the larger saving is fewer discarded attempts.
There is a second test that matters especially for brands, and it is the one Google’s reference-adherence claims are aimed at. A product video fails commercially if a face changes between angles, packaging text drifts, or a brand colour looks different from one shot to the next, even when the clip is technically impressive. Small studios and social teams in the United States will therefore judge Omni 1.2 Flash less by its best demonstration than by whether ordinary product shots stay recognisable across a whole campaign. The SynthID watermark adds a different kind of practical value for those businesses, giving AI-generated advertising an invisible marker as disclosure expectations grow. If consistency and editing hold up together, lower-priced generation changes who can afford regular video advertising; if consistency still breaks on close inspection, the low price mainly buys faster drafting before a conventional production process takes over.
Read Next
- Tencent’s Hy4 Can See and Use a Computer — but Its Open Weights Are Not Out Yet
- Apple Cuts Engineers’ Daily AI Budget From $300 to $200 as Token Bills Bite
Sources
