Gemini Omni 1.2 Flash: a 15-second clip now costs about what 10 seconds did
Google's new video model cuts the price 30% and stretches clips to 15 seconds. For developers, the per-second math is what changes.
Gemini Omni 1.2 Flash makes a full 15-second clip cost about what a 10-second clip cost before. At $0.07 a second for 720p, a 15-second clip runs $1.05. Google says the new price is 30% below Omni 1.1. That puts the old rate near $0.10 a second, which is my arithmetic, not a figure Google published. On that math, a 10-second clip used to cost about $1.00.
Google’s announcement, dated November 5, was written by Anish Nangia, a product manager at Google DeepMind. Google’s blog carries the same text, so the two links are one announcement, not two confirmations.
The math a developer should run first
Omni 1.1 topped out at 10 seconds. Omni 1.2 Flash goes to 15, and you can pick any length from 3 to 15 seconds in one-second steps. Pricing is quoted per second at 720p, so the length you request sets the bill. Ask for six seconds and you pay for six.
| Omni 1.1 | Omni 1.2 Flash | |
|---|---|---|
| Maximum clip length | 10 seconds | 15 seconds |
| Duration choice | Not in the announcement | 3 to 15 seconds, 1-second steps |
| Price at 720p | About $0.10 a second (my math) | $0.07 a second |
| Full-length 720p clip | About $1.00 for 10 seconds (my math) | $1.05 for 15 seconds |
A 10-second shot now costs $0.70, down from about $1.00. Those five extra seconds add $0.35. My read: a team already cutting its shots to 10 seconds gets the bigger benefit from the price drop. The extra length only helps if your shots were getting cut short.
What the draft-then-upscale path does to the bill
Google’s suggested workflow is to draft at 360p or 720p, fix details with a follow-up edit instead of regenerating the whole clip, then upscale the keepers to 1080p or 4K. Google presents that as the cost-saving path. It doesn’t give a per-second price for 1080p, for 4K, or for the upscale step. Only the $0.07 figure appears as a per-second rate in either announcement.
Google’s per-second rate covers 720p only.
My read: the upscale step is where a budget can go wrong, and the post leaves it unpriced. Before committing a batch, run one clip through your full draft-and-upscale path and log what it costs.
Google’s Gemini API pricing page lists Omni 1.2 Flash by token instead. It shows $1.50 per million input tokens for text, image or video, and $12 per million output tokens for video. The same page lists $17.50 per million video output tokens for Omni 1.1 Flash. I did not convert those token rates into seconds. Google’s per-second figure and its token rates should agree, but I haven’t checked that against a real billing record.
What the announcement doesn’t cover
- Competing video APIs. I did not pull current prices for other services, so I can’t say whether $0.07 a second is cheap in the market.
- Speed. The Google blog summary says “faster” generation, but the post gives no generation times.
- Quality. Consistency and lip-sync examples are Google’s own clips, picked by Google. I did not run the model.
- Known gaps. Google says work continues on fast sports action, rapid script changes and complex physics. For ad spots with quick cuts, that matters.
Where else it shows up
Google says Omni 1.2 Flash is also in Flow, in Google Vids at no extra cost for 15-second clips (with a Google account or Workspace subscription), in the Gemini app (rolling out to AI Ultra subscribers), and in Google Ads. Enterprise customers get Provisioned Throughput, and outputs carry a SynthID watermark.
Gemini 4 Argon raised the same kind of question, where the cheaper bill depended on a launch discount. See Gemini 4 Argon ties GPT-6 Astra for 60% of the cost.
My read: the 30% cut matters most for the uses Google’s own examples show: ad variants, catalog spins and virtual try-on. Those are short, high-volume and repeated, so a lower per-second rate multiplies. A one-off clip barely changes.
Note on sources: I read Google DeepMind’s post and Google’s blog post in full, and the Omni rows of Google’s AI Studio pricing page. I did not run the model, call the API, or open the Google Cloud availability post, or the Flow, Vids and Ads pages. Google’s 30% figure is the only source for the cut. My $0.10 Omni 1.1 rate and $1.00 comparison are arithmetic from that figure.