# Google Cuts Video AI Costs 30% With Gemini Omni 1.2 Flash GA Rollout

> Google DeepMind's Flash-tier video model reaches production with native 15-second clips, per-second pricing, and SynthID watermarking for high-volume commercial workflows.

*Published 2026-11-08 · By Marcus Vance*

Gemini Omni 1.2 Flash is Google's generally available multimodal model for native video generation and editing, priced at $0.07 per second for 720p output and supporting clips up to 15 seconds long.

Google has moved Gemini Omni 1.2 Flash from preview to general availability on the Gemini API and Gemini for business, giving developers and enterprise customers a production-ready model for short-form video generation and editing. The release extends native video clips from 10 seconds to 15 seconds and cuts the price of 720p video output by 30%, to $0.07 per second.

Product managers at Google Cloud and Google DeepMind positioned the release as a response to high-volume commercial use cases rather than cinematic filmmaking. The announcements emphasize product videos, virtual try-ons, and ads, and the per-second pricing structure is designed to make those workloads predictable at scale.

## What does the Gemini Omni 1.2 Flash rollout include?

Gemini Omni 1.2 Flash is now generally available on the Gemini API and through Gemini for business, according to Google. Native video generation supports clips up to 15 seconds long, up from 10 seconds in the previous version, and users can request exact durations in one-second increments from 3 to 15 seconds.

Google also says the model supports multi-turn conversational editing across full clips. That capability lets a user generate a video, request a change, and continue iterating without regenerating the entire clip from scratch. The workflow is intended to fit the iterative production loops common in marketing and commerce teams.

Outputs carry native audio and are watermarked with SynthID, Google's provenance tool for AI-generated content. Audio references require allowlisting, which means organizations must be approved by Google before they can supply reference audio to steer generation.

## How does the new pricing compare with prior versions?

Google sets the list price for 720p video output at $0.07 per second, a 30% drop from Gemini Omni 1.1 Flash. At that rate, a 15-second clip costs $1.05 in 720p output before any upscaling. The reduction targets teams producing video in volume, where per-second cost is a primary driver of tooling choices.

| Capability | Gemini Omni 1.2 Flash |
| --- | --- |
| Availability | Gemini API and Gemini for business |
| Native video clip length | Up to 15 seconds, from 3 to 15 in one-second increments |
| 720p price per second | $0.07 |
| Price change vs. Gemini Omni 1.1 Flash | 30% lower |
| Draft resolutions | 360p or 720p |
| Upscale targets | 1080p or 4K |
| Editing | Multi-turn conversational editing across full clips |
| Audio | Native audio output; audio references require allowlisting |
| Provenance | SynthID watermark on outputs |

The price cut is notable because video generation carries higher compute demand than text or image generation, and Google is using a Flash-tier model to signal that routine commercial video can be treated as a commodity workload. The decision to offer per-second billing, rather than per-clip pricing, aligns cost with the exact length of the requested output.

## What technical capabilities distinguish Gemini Omni 1.2 Flash?

Google lists five technical changes that define the update from the prior version.

- Native generation and editing of video clips up to 15 seconds, an increase from 10 seconds in the prior version.
- One-second duration control from 3 to 15 seconds, giving developers precise control over output length and cost.
- Draft generation at 360p or 720p with upscaling to 1080p or 4K.
- Multi-turn conversational editing that can apply changes across full clips.
- Native audio output with SynthID watermarking; audio reference inputs require allowlisting.

Duration control in one-second increments is a practical differentiator for programmatic use. An advertising platform that needs a 6-second bumper, a 12-second product demo, or a 15-second social cut can request the exact length and pay only for that length.

Multi-turn conversational editing also changes the production workflow. Rather than generating multiple candidate clips and selecting the best one, a user can generate a draft and then direct changes through natural language. Google says the editing applies across the full clip, not just isolated frames.

Upscaling to 1080p or 4K lets teams keep draft iteration cheap at lower resolutions and reserve high-resolution output for the final stage. The model produces native audio, which removes a separate audio-generation step, and watermarks generated content with SynthID to preserve provenance.

## How will lower video generation costs affect enterprises and developers?

The commercial framing is explicit. John Nack, senior product manager at Google Cloud, said the model is built for the everyday videos businesses make in volume, including product videos, virtual try-ons, and ads. That framing puts Gemini Omni 1.2 Flash in competition with a range of short-form video generation services aimed at marketing and commerce teams.

For businesses, per-second pricing makes budget calculations straightforward. A marketing team producing 100 15-second 720p ads would spend $105 on model output at list price, before upscaling and platform fees. The 30% price reduction effectively increases the volume of video a fixed budget can buy, which matters for A/B testing and personalized creative at scale.

The model is distributed through two channels: the Gemini API for developers building video applications and Gemini for business for enterprise customers. The dual path means the capability is available both to independent developers and to large organizations standardizing on Google's productivity stack.

The 30% cut also signals competitive intent. By publishing a lower per-second rate for 720p output, Google is telling the market that short commercial video should be priced as a high-volume utility rather than a premium creative service.

## What are the safety and provenance controls?

Google is shipping provenance controls alongside the creative features. Generated video carries a SynthID watermark, which embeds a detectable marker into the output so platforms and viewers can identify AI-generated content. Native audio is included in the output, and audio references are restricted to allowlisted organizations.

The allowlisting requirement for audio references suggests Google is taking a cautious approach to voice cloning and sound-alike use cases. By requiring approval before organizations can supply reference audio, Google retains a control point over a capability that carries impersonation risk. The company did not detail the approval criteria in the GA announcement.

## What constraints should developers plan around?

The model's 15-second ceiling is the most obvious constraint. Teams building longer narratives, documentaries, or multi-scene spots will still need to composite multiple clips or use a different model. The one-second increment control mitigates that limitation for short-format work, but it does not remove the ceiling.

Audio reference inputs are another gating point. Because audio references require allowlisting, an organization cannot simply upload a voice sample or music reference on day one; it must be approved by Google first. Teams should factor that approval process into launch timelines if their use case depends on steering output with audio.

Resolution also carries an implicit workflow. Drafts can be generated at 360p or 720p and upscaled to 1080p or 4K, so teams that need final high-resolution assets should plan a two-stage pipeline: iterate at draft resolution, then upscale the approved clip. Google has not published separate pricing for upscaling in the GA announcement, so cost estimates should treat upscaling as an additional step.

## What does general availability mean for production use?

General availability status is a meaningful threshold for enterprise adoption. It signals that Google is committing to the model's API stability, support, and billing terms, and it removes the expectations that come with preview releases. For teams that were piloting Gemini Omni 1.1 Flash, the GA release gives them a clear production target with a lower published price.

Distribution through the Gemini API and Gemini for business also reduces the integration burden. Developers who already use Google's developer tools can call the model through a familiar interface, while business customers can access it inside Google's commercial product family without building a separate pipeline.

## What did Google executives say about the release?

Google DeepMind product manager Anish Nangia tied the release to both capability and cost in the company's announcement. 'Gemini Omni 1.2 Flash introduces native 15-second video creation, multi-turn conversational editing capabilities, and enhanced reference following,' Nangia said. 'We have dropped the price 30% from Gemini Omni 1.1 making it available at $0.07 per second for 720p video.'

> Today, Gemini Omni 1.2 Flash is generally available on Google Cloud. Built for the everyday videos businesses make in volume — like product videos, virtual try-ons, and ads — this model is fast, cost-effective, and follows your image, video and audio references more closely than before.John Nack, Senior Product Manager, Google Cloud

Nangia's emphasis on enhanced reference following points to the model's ability to maintain consistency with supplied images, video clips, and audio. That matters for branded content, where a product's appearance and the style of prior creative assets must carry through multiple generations.

## How does Gemini Omni 1.2 Flash fit into Google's broader AI strategy?

Gemini Omni 1.2 Flash is a Flash-tier model, and the GA announcements consistently pair speed and cost-effectiveness with creative capability. That pairing reflects a broader strategy: Google is trying to make video generation feel as routine as image generation, with predictable pricing and provenance built in from the start.

The decision to include native audio and SynthID watermarking in the same output also signals that Google wants safety to be part of the default production path, not an add-on. Developers who build on the Gemini API inherit those controls rather than assembling them separately.

The market effect is likely to be felt most in advertising technology and e-commerce, where the cost of producing many short video variants has historically limited experimentation. At $0.07 per second for 720p, the economics of generating and testing a large set of creative variants change materially.

## What comes next for Google's video generation lineup?

With Gemini Omni 1.2 Flash in general availability, Google's near-term roadmap appears focused on volume, iteration speed, and cost rather than longer single clips. The 15-second ceiling covers a large share of social and advertising formats, while the one-second increment control and multi-turn editing reduce the friction of producing those clips in bulk.

The combination of a Flash-tier model, per-second pricing, and enterprise distribution suggests Google is treating short video as an infrastructure product. Future updates could extend clip length, improve reference fidelity, or add more granular control over audio, but Google has not announced a roadmap for the next Omni version.

For now, the practical benchmark is straightforward: a 15-second 720p clip costs $1.05 on the Gemini API, and the full output includes native audio and a SynthID watermark. That combination of length, price, and provenance gives developers a defensible basis for comparing Gemini Omni 1.2 Flash with other short-form video models.

Google's general availability rollout signals that the competitive frontier in video AI has shifted from raw capability demos to cost per useful minute. The company is betting that predictable per-second pricing, shorter clips, and provenance controls will win the high-volume commercial market even as longer-form models continue to improve.

## Sources

1. [Gemini Omni 1.2 Flash is generally available on the Gemini API and Gemini for business; supports native 15-second video creation; priced at $0.07 per second for 720p, 30% lower than Gemini Omni 1.1 Flash.](https://blog.google/innovation-and-ai/technology/developers-tools/introducing-gemini-omni-1-2-flash/)
2. [Gemini Omni 1.2 Flash is generally available on Google Cloud for high-volume business video; priced at $0.07 per second of 720p video, 30% less than Gemini Omni 1.1 Flash.](https://cloud.google.com/blog/products/ai-machine-learning/gemini-omni-1-2-flash-generally-available)
3. [Gemini Omni 1.2 Flash generally available on Gemini API; supports up to 15-second video clips at 30% lower pricing than prior version ($0.07 per second for 720p).](https://theagenticweekly.com/news/models)

---
Source: https://aiintelreport.com/frontier-models/gemini-omni-1-2-flash-ga-video-pricing
Index: https://aiintelreport.com/llms.txt · Full text: https://aiintelreport.com/llms-full.txt