Google cut the price of its AI video model on Nov. 51. Gemini Omni 1.2 Flash now turns a second of 720p video into 7 cents, 30 percent less than the version it replaces3. Seven production companies already route real work through it, Google said the same day.
Clips also run longer now. A single video stretches to 15 seconds, up from 10, in one-second steps2. invideo tested the new model before launch. Its engineers called it one of the fastest video models the company has run. A review later in this piece checks that claim.
The Price Drops, the Clip Grows
Google DeepMind built the earlier version for 10 cents, or $0.10, a second. That was Gemini Omni 1.1 Flash, priced per second of 720p video3. It held that price since August.
The new version cuts that to 7 cents, or $0.07. It also lets one clip run half again as long, from three to fifteen seconds in one-second steps2.
Google Cuts Gemini Omni's Video Price 30 Percent to 7 Cents
Price per second of 720p video, Gemini Omni Flash, August to November 2026
Each column is the price Google charged for one second of 720p video; the November column is 30 percent lower than the August one.
Sources: Google Cloud Blog [1]; The Decoder [3]. Chart by Kelly Dennis.
The numbers behind this chart
| Item | Value |
|---|---|
| Omni 1.1 (Aug.) | $0.10 |
| Omni 1.2 (Nov.) | $0.07 |
Builders reach the model four ways today. Developers pay by the second through Google Cloud's console and the Gemini API. Its video editor, Vids, gives the model free to anyone with a Google account or a Workspace plan. The same model is also rolling out inside the Gemini app, for its AI Ultra subscribers2.
Every clip still carries an invisible SynthID mark, Google's watermark for flagging AI video2. Google says the mark now ships by default on every clip this model makes. That matches the promise it made for earlier video tools.
Large customers can also buy Provisioned Throughput on Google Cloud. The purchase reserves a lane of rendering capacity to keep a busy studio's queue moving during its own launch week.
Seven Companies Call It Production Ready
Google published quotes from seven partners the same day it cut the price. Shivam Rastogi, vice president of engineering at invideo, ran the new model through his team's own tests before launch. "We ran Gemini Omni 1.2 Flash through our evals at invideo. Honestly, it's really good. It sticks to the prompt, and the physics and sound feel real," Rastogi said1.
Matt Rouif, chief executive and co-founder of Photoroom, already ships the model inside a paying product. "Photoroom turns product photos into video ads that sell in minutes. We chose Gemini Omni Flash to power our Video Ad Maker and Product Animator because it balances quality, speed and cost," Rouif said1.
Richard Girges, chief technology officer at QuickFrame AI, builds ads for connected television with the same model. "QuickFrame AI helps marketers and business owners create, edit, and launch professional-quality ads for CTV and social in minutes. With Gemini Omni 1.2 Flash now supporting 15-second video generations at 1080p, users can create a complete CTV ad in a single generation," Girges said1.
Narek Hayrapetyan, head of product and partnerships at Picsart, plans to push the model toward a far bigger crowd. "Gemini Omni 1.2 fits the way our teams work. Draft mode lets us iterate concepts in minutes, then render in 4K for TV-ready AI dramas," Hayrapetyan said. Picsart counts more than 100 million creators on its own platform1.
Google for Developers@googledevsMeet Gemini Omni 1.2 Flash ⚡️
Experience higher quality and consistency, native 15-second video generation and editing, and 30% lower pricing to help you build production-ready outputs across the @Google ecosystem. Check out the model updates 🧵
A Review Routes the Hardest Jobs Elsewhere
Rohit Rao founded Segmind, a company that tests AI media models against each other. He published his own hands-on read of Gemini Omni 1.2 Flash the day it launched. The test pitted it directly against Seedance, a rival video model4. ByteDance built Seedance and sells it through its own cloud tools6.
"Identity holds across shots, but motion reads floaty and faces sit below Seedance," Rao wrote4. He found a second gap in how closely the model follows a prompt. "About one instruction in four gets skipped, usually a background change or a final camera move," Rao wrote. That split works out to a 75 percent hit rate against a 25 percent miss rate on his own test.
Segmind's Review Finds Gemini Omni Follows 3 of 4 Instructions
Share of prompt instructions followed in Segmind's hands-on review of Gemini Omni 1.2 Flash, Nov. 5, 2026
The colored arc is the share of instructions the review found the model followed; the gray share is the one in four it skipped, usually a background change or a final camera move.
Source: Segmind [4]. Chart by Kelly Dennis.
The numbers behind this chart
| Item | Value |
|---|---|
| Instructions followed | 75% |
| Instructions skipped | 25% |
| Whole | 100% |
Rao called the model obedient over pretty. He routed his own jobs by that line. Fast drafts and background swaps went to Gemini Omni.
Photoreal people, lip-synced acting and a brand's hero ad went to Seedance instead4. invideo and Photoroom build exactly the kind of fast, high-volume work Rao still sends to Google's model. Both companies kept shipping on it anyway.
What to do
Route by job, ahead of headline price. Send drafts, background swaps and high-volume ad variants to Gemini Omni 1.2 Flash at 7 cents a second, as invideo and Photoroom do14.
Keep a brand's hero shot, a lip-synced spokesperson or anything resting on a human face on a different model for now. Segmind's own review still rates that other model higher, until a second independent test says otherwise4.
Try Google Vids first on a tight budget. Any Google account or Workspace plan already includes the new model free2.
By the numbers
- 7 cents: price Google charges per second of 720p video on Gemini Omni 1.2 Flash, down from 10 cents14.
- 30 percent: the size of that price cut, Google's own figure for the new release1.
- 15 seconds: the longest native clip the model generates in one pass, up from 10 seconds in the prior version2.
- 1 in 4: prompt instructions skipped, per Segmind's review, usually a background change or a final camera move4.
- 7: production companies, from invideo to Picsart, that published quotes calling the new model ready for live work1.
- 100 million: creators Picsart counts on its own platform, where the model is rolling out1.
What to watch
Watch whether a second lab runs its test against Seedance and lands near Segmind's split verdict, or closer to Google's partners. The Gemini app's rollout to Google AI Ultra subscribers is also worth tracking. A free tier changes who tries the model first2.
This model still has to earn its place on tougher jobs too. That happens when a marketer picks it over a pricier rival for work that matters. Marketers will settle that question, ahead of any blog post.
Sources
- "Gemini Omni 1.2 Flash now GA: cost-effective video," Google Cloud Blog, Nov. 5, 2026, https://cloud.google.com/blog/products/ai-machine-learning/gemini-omni-1-2-flash-generally-available
- Anish Nangia, "Gemini Omni 1.2 Flash: 15-second video and lower pricing," Google, Nov. 5, 2026, https://blog.google/innovation-and-ai/technology/developers-tools/introducing-gemini-omni-1-2-flash/
- "Google's Gemini Omni 1.1 Flash makes AI video generation cheaper and more flexible," The Decoder, Aug. 27, 2026, https://the-decoder.com/googles-gemini-omni-1-1-flash-makes-ai-video-generation-cheaper-and-more-flexible/
- Rohit Rao, "Gemini Omni 1.2 Flash: Google's New Gemini Video API Tested," Segmind, Nov. 5, 2026, https://blog.segmind.com/gemini-omni-1-2-flash-review/
- Google for Developers (@googledevs), post on X, Nov. 5, 2026, https://x.com/googledevs/status/2118389720111866253
- "Seedance," Wikipedia, accessed Nov. 7, 2026, https://en.wikipedia.org/wiki/Seedance
