Google ships Gemini Omni 1.2 Flash to GA — 15-second clips, full-clip editing, 30% cheaper
Google rolled out Gemini Omni 1.2 Flash on November 5, the first model in the Gemini Omni family to reach general availability. It generates and edits video up to 15 seconds — up from 10 — with multi-turn conversational editing across the full clip, and it costs 30% less at $0.07 per second for 720p video. Google published no benchmarks for the quality claims — those are vendor claims, not results.
📊 Scorecard: our Bullish call on $GOOGL is a miss so far: from $288.15 to $287.38 at the latest close (-0.3%). Against QQQ the call is right: the stock is 0.4 points ahead of the index.
The News
Google rolled out Gemini Omni 1.2 Flash on November 5, the first model in the Gemini Omni family to reach general availability — the stage where a product is fully released rather than in preview. It is available across the Gemini API, Google AI Studio, Google Cloud with Provisioned Throughput, and Gemini for business, plus Flow, Vids, the Gemini app and Ads. In the Gemini app, 15-second clips are rolling out to AI Ultra subscribers. General availability matters because it is the point at which businesses can build on a model with confidence it will stay supported.
The headline changes versus Omni 1.1 are length, editing and price. Native video generation and editing now run up to 15 seconds, up from 10, with duration settable in one-second steps from 3 to 15. Multi-turn conversational editing works across the full clip — swap a product or background, change lighting or weather, switch aspect ratio — with stronger first-to-last-frame consistency, first/last keyframe interpolation and clip extension. Speech is more natural, with clearer lip-sync and multi-speaker conditioning, which means the model can handle scenes where more than one person talks. Outputs carry an imperceptible SynthID watermark, Google's invisible marker that identifies video as AI-generated.
The price cut is the commercial news: 30% cheaper, at $0.07 per second for 720p video, or $12.00 per million video output tokens. Older models retire on a schedule — Veo Extend was retired on November 5, and in Flow, Veo 3.1 Fast, Veo 3.1 Quality and Omni Flash 1.1 retire on December 5; on the API, requests to Omni 1.1 auto-route to 1.2 from December 3. One limit to note: Google published no benchmarks for the quality claims, so the improvements in consistency and speech are vendor claims, not independently verified results.
Longer clips, full-clip conversational editing and a 30% price cut — Google's video model goes general, with the quality claims still unproven.
Why It Matters
This matters because video is where AI generation is becoming a real business, and the constraint has been length and cost. A 10-second clip is a teaser; a 15-second clip with editing across the whole thing is a usable ad, a product demo or a social post. The one-second steps from 3 to 15 seconds matter, too — advertisers buy video by the second, and being able to dial the exact length is the difference between a model that fits a workflow and one that does not.
It matters for Google's position, because Omni 1.2 is now the default across its surfaces. Flow, Vids, the Gemini app and Ads all run on it, and the December retirements push every existing user onto the new model. That is the kind of distribution no standalone video startup can match: Google is not just shipping a model, it is shipping it to every customer it already has. The 30% price cut reinforces that — at $0.07 a second, the cost of experimenting falls to the point where teams try video generation for jobs they would never have considered before.
And it matters with its limits stated. The quality claims — better consistency, more natural speech, multi-speaker conditioning — come from Google, with no benchmarks published. Until independent tests exist, the honest read is that the length and the price are verified and the quality is promised. But length and price are the two things a buyer can check immediately, and on both, this release delivers.
Analyst's Take
Our call is Bullish. This is a general-availability release, not a demo — and the combination of longer clips, conversational editing across the full clip and a 30% price cut is the kind of package that moves video generation from experiment to production. The 15-second ceiling, with one-second steps from 3 to 15, is long enough for a real ad, a product clip or a social video. The price — $0.07 a second — makes it cheap enough to iterate. For Google, the point is the funnel: Flow, Vids, the Gemini app and Ads all run on the same model, so every surface is now selling the same capability.
What to watch: whether anyone publishes independent benchmarks, because Google's quality claims — consistency, lip-sync, multi-speaker speech — are still vendor claims. The December 3 auto-route and December 5 retirements force existing users onto 1.2 whether they like it or not, which is a vote of confidence from Google and a deadline for everyone else. If the quality holds up in real use, the 30% cut is the number that matters most: video generation is a volume business, and the cheapest capable model tends to win the volume.
Are you bullish or bearish on this story?
By The Numbers
FAQ
What is Gemini Omni 1.2 Flash, in plain terms?
It is Google's AI model for making and editing video from text and image prompts. You describe a clip or upload one, and the model generates or changes it — swapping a product or background, changing the lighting or weather, or switching the aspect ratio. Flash is the fast, cheaper tier of the model family.
What changed versus Omni 1.1?
Clips run up to 15 seconds, up from 10, with duration settable in one-second steps from 3 to 15. You can now edit the whole clip in a multi-turn conversation instead of one shot at a time, with stronger first-to-last-frame consistency, first/last keyframe interpolation and clip extension. Speech is more natural, with clearer lip-sync and multi-speaker conditioning.
What does it cost?
$0.07 per second for 720p video — a 30% price cut — or $12.00 per million video output tokens. It is available across the Gemini API, Google AI Studio, Google Cloud with Provisioned Throughput, Gemini for business, Flow, Vids, the Gemini app and Ads. In the Gemini app, 15-second clips are rolling out to AI Ultra subscribers.
What is SynthID?
It is Google's watermarking system for AI-generated content. Omni 1.2 outputs carry an imperceptible SynthID watermark — invisible to the eye, but detectable by software — so generated video can be identified as AI-made.
What is being retired, and when?
Veo Extend was retired on November 5. In Flow, Veo 3.1 Fast, Veo 3.1 Quality and Omni Flash 1.1 retire on December 5; on the API, requests to Omni 1.1 auto-route to 1.2 from December 3. If you use an older model, you need to move to 1.2 by then.
👁
Source: Google