FervorCreative AI
Live Latest 07.11.26 · morning 331 tools tracked 824 workflows indexed 533 topics Hot: Claude Code, Seedance 2.5, MiniMax H3

Gemini Omni 1.2 Flash makes a 15-second clip you can fix by asking instead of rerolling, and it is free inside Google Vids, but Vids clips are meant to stay in Vids, so the smart move is to sketch for free and pay the API's roughly one dollar per 15-second draft for anything you will deliver.

Gemini Omni 1.2 FlashGoogle VidsGoogle FlowGemini APISynthIDvideo-genai-editingcreative-workflowslicensing-provenance

Gemini Omni 1.2 Flash Gives You Free 15-Second AI Video in Google Vids. Read the Fine Print Before You Plan a Client Job Around It

Fifteen seconds is a strange number to get excited about. It is shorter than most songs' intros. But it is also the length of a lot of paid work: a pre-roll ad, a product loop on a shop page, a social teaser, the cold open of a short. For most of the last year, AI video models stopped well short of that, and you stitched clips together and hoped the seams would not show.

Google's new Gemini Omni 1.2 Flash, released on November 5, makes the full 15 seconds in one go. It also lets you fix a clip by asking for a change, instead of throwing it away and starting over. And Google is giving it away inside Google Vids, its slideshow-style video app: you can "generate 15-second clips at no additional cost in Google Vids with a Google account or Google Workspace subscription."

That last part is the headline everyone will repeat. It is also the part worth reading twice.

My position is simple. Omni 1.2 Flash is a real step for anyone who makes short video, and the free version is a great sketchpad. It is not a free production tool. Treat Vids as the place you figure out the shot, and budget for the paid version when the clip has to leave the building.

What actually changed

Three things in Google's announcement matter to someone who edits for a living.

Clips are 15 seconds now. The previous version topped out at 10. You can set the length in one-second steps, anywhere from 3 to 15 seconds. That sounds small, but five extra seconds is the difference between a shot and a sequence. A product can enter, turn, and settle. A character can say a full line and react.

You can fix a clip instead of rerolling it. Google calls this multi-turn editing, which just means a back-and-forth conversation about the same clip. Its examples: "swapping a product or background, changing the lighting or weather, or switching the aspect ratio." One demo keeps the same shot while the background moves from a striped tent to an oak forest, a stable with a white horse, a market fair and a stone castle. Anyone who has burned an afternoon rerolling a nearly perfect clip because one detail was wrong will understand why this matters more than the extra length.

You draft small and finish big. Google's suggested routine is to "draft fast at 360p or 720p, fix a detail with a follow-up edit instead of re-generating the entire clip, and then upscale your best creations to 1080p or 4K." That is how photographers have always worked with contact sheets. Look at a lot of cheap versions, then spend real effort on the keeper.

Sound comes with the picture. Google says dialogue is "more fluid" and lip-sync is clearer, and it can keep several speakers distinct. You can also feed it references: images of a character, product or style, short motion clips "up to three seconds," audio cues, and one-second voice samples.

Google is candid about the weak spots too. Work continues on "fast sports action, rapid script changes and complex physics." If your shot involves a skateboard trick or a glass of water being knocked over, expect to fight it.

Where it lives, and what each door costs

Omni 1.2 Flash shows up in several Google products at once, and they are not equal.

Google Vids is the free door. Vids is Google's video editor for presentations, explainers and team updates, and it now generates 15-second clips at no extra charge. Google's November 5 Workspace update lists personal Google accounts among those who get it, and says people with the AI Expanded Access add-on get higher limits.

Flow is Google's filmmaking app, built for directing scenes and multi-character dialogue. Google says you can "direct scenes and multi-turn edits in Flow," but did not publish Flow's credit cost for the new model in the announcement.

The Gemini app gets it too, "rolling out to Google AI Ultra subscribers," Google's top paid plan.

The Gemini API is the paid door for developers and studios, priced at "$0.07 per second for 720p video," which Google says is 30% cheaper than the last version. A full 15-second draft at 720p works out to about $1.05. Ten drafts of a shot cost about ten dollars.

One caution on that number. When I checked the Gemini API's own pricing page the day after launch, it still listed only the older Omni 1.1 Flash, at about $0.10 per second for 720p, with no free tier and no listed price for upscaling. Google's blog may simply be ahead of its docs. Check the price in Google AI Studio before you quote a client.

The fine print on the free version

Here is the sentence that changes how I would use Vids. Google's help page for generating clips in Vids says: "Generated images and videos are for use only within Vids."

That page also says "most users can generate up to 50 videos per month," with the limit resetting on the first of each month, and it still describes clips of 3 to 10 seconds. So it may be out of date, and Google may update both the limit and the wording. But as written today, the free clips are meant to stay inside Vids projects. That fits Vids' purpose (internal explainers, training videos, team updates) and it does not fit "download it and drop it into a client's Premiere timeline."

Compare the developer terms for the Gemini API. They say "Google won't claim ownership over that content," and that "you're responsible for your use of generated content." That is the language of a production tool. The Vids language is the language of a feature inside an office app.

Every clip from every door also carries an "imperceptible SynthID watermark," Google's invisible tag that marks the video as AI-made. You cannot see it, and it travels with the file. If a client has a policy on disclosing AI footage, this is the version of the truth that travels with the file.

None of this is a scandal. It is just the difference between free and paid that most launch coverage skipped.

Put this into practice

Here is how I would fold Omni 1.2 Flash into real work this week.

1. Use Vids as your storyboard. Open Google Vids and block out the shot in rough clips. Try the opening, the camera move, the timing of the product reveal. You are not making the final asset here. You are deciding what the final asset is, for free, and Google's help page lists up to 50 generations a month for most users.

2. Write the edit list, not a new prompt. When a clip is close, resist the reroll. Ask for the specific change in plain words: "same shot, change the background to a rainy street," "same shot, warmer late-afternoon light," "make this vertical." Keep a running note of the edits that worked, because that note becomes your recipe for the paid render.

3. Lock references before you pay. Gather the character image, the product photo, any short motion clip, and a one-second voice sample if someone speaks. The more of the look you pin down with references, the fewer paid drafts you burn.

4. Move the keeper to the API or Flow. Once you know the shot, generate it through a paid route, at 720p first. At about a dollar per 15-second draft, budget for three to five drafts per shot, so $3 to $5, and up to ten dollars if the shot is fiddly.

5. Upscale only the winner. Draft at 360p or 720p, then upscale the take you will actually use to 1080p or 4K. Check the price of that step when you do it, because Google had not published one at launch.

6. Keep the edit on your own timeline. Bring the final clip into your editor and do the cutting, the grade and the sound there. Omni can swap a background, but your timeline is where you can see every decision.

Honest limitations

The free tier is a sketchpad, not a delivery tool. Vids clips are, per Google's own help page, "for use only within Vids." If the work leaves Vids, use a paid route.

The prices are a moving target. The blog says $0.07 a second. The API pricing page said about $0.10 a second for the older model when I checked, and nothing for the new one or for upscaling. Confirm before you quote.

Physics still breaks. Google itself names fast sports action, rapid script changes and complex physics as unfinished. A shot full of quick motion will need more drafts, or a different approach.

Every clip is tagged. The SynthID watermark is invisible and built into every clip. That is good for honesty and worth knowing before you hand footage to a client who has rules about AI.

Fifteen seconds is still fifteen seconds. Longer pieces still mean joining clips, and matching the look across joins is still your job.

The real gift is the edit, not the price

The free clips will get the headlines. I think the bigger change is that you can now keep a nearly right clip and fix it, the way you would give a note to a camera operator, instead of throwing it out. That turns AI video from a slot machine into something closer to a conversation with a crew.

So take the free 15 seconds for what it is: the cheapest storyboard you have ever had. Rough out your next short spot in Vids this weekend. Write down the edits that made it work. Then decide, with real numbers in front of you, whether the finished version is worth a few dollars a shot. For most of the work I see, it will be.


From the briefing