Google Just Made AI Video Cost Three Cents a Second. Do the Math.

Last week, a client asked me what a 30-second product demo video would cost. The freelancer quote came back at $2,400. Two-week turnaround.

Then Google dropped Gemini Omni 1.1 Flash on August 27. That same 30-second clip runs $0.90 at draft quality. A polished 1080p version: $4.50.

I checked the math twice. It's correct.

Look at the pricing structure

Google didn't go with per-video or monthly subscription pricing. They charge per second across four resolution tiers:

• 360p drafts: $0.03/second
• 720p: $0.10/second
• 1080p: $0.15/second
• 4K: $0.30/second

This matters because you can test ideas cheaply at low resolution, then upscale only the clips that actually work. You're not spending $0.30 per second on concepts that might flop.

What shipped beyond the price tag

The model added features that make business video production practical, not just cheap.

Scene extension generates clips up to 40 seconds long. That covers a social ad, a product walkthrough intro, or a training segment. Keyframe control lets you lock start and end frames so the footage goes exactly where you need it. And video references keep characters and settings consistent across multiple clips.

That consistency piece matters most. Every AI video tool so far has struggled here. You'd get a great 10-second clip, then the next one would look like it came from a different production entirely. Google seems to have made real progress on this.

Who should pay attention

If your business produces any of these, run the numbers on what you're currently spending:

Social media video ads. Product demos. Internal training content. Real estate walkthroughs. Client presentation intros.

A small marketing agency producing 20 social clips per month at 15 seconds each could generate 1080p drafts for about $45 total. Throw away half, regenerate, and you're still under $100. Compare that to stock footage licensing or freelance editing invoices.

I'd pump the brakes before replacing your video team, though. AI video still stumbles on specific brand elements, precise text overlays, and anything requiring exact product placement. It's a drafting tool that needs a human finishing pass.

Iteration speed is the bigger deal

I think most people will focus on the price and miss what really changed.

Video has always been the most expensive content format to iterate on. You shoot, edit, review, reshoot. Each cycle costs time and money, so most businesses cap themselves at one or two revision rounds.

At three cents per second for drafts, you can generate 50 variations of a concept before lunch. Test different openings. Try three visual styles. Figure out which hook grabs attention in the first two seconds. Then take the winner and upscale it.

That's how good text marketers already work. They write 20 headlines and pick the strongest one. Video has been too expensive for that kind of volume testing. Not anymore.

What to try this week

The model is live in Google AI Studio and the Gemini Developer API right now.

Take your most recent marketing video and try to recreate the concept with Gemini Omni 1.1 Flash. Don't aim for a perfect match. See how close you can get, and note where the gaps show up. That gives you a clear map of where AI video helps your workflow and where it falls short.

The agencies I work with are already testing a hybrid setup. AI generates the base footage and rough cuts. Your team focuses on the stuff that requires taste: branding, pacing, color, and polish. Generating raw footage drops to pennies. The creative judgment stays with the people who know your brand.

Video production just went from a $2,500 commitment to a $5 experiment. The cost of finding out whether AI video works for you is now pocket change.

— Mark Garza, Laimen AI