Foreword¶

Once multimodal video generation capabilities enter the API and enterprise product channels, engineering teams care about three things: how long a single generation can cover, how many rounds of interaction and computing power are needed for a revision, and whether the price list can withstand mass production. Google DeepMind announced the launch of Gemini Omni 1.2 Flash on November 5, 2026, in an official blog post, making it generally available (GA) on the Gemini API and Gemini for Business, while also providing longer native video durations and clearly reduced per-second pricing for developers and enterprise workflows. Below, I break down the release notes item by item to help evaluate whether it is worth integrating into existing pipelines.

Release Date and Availability¶

The blog post is signed by Google DeepMind Product Manager Anish Nangia, with a release date of November 5, 2026. The text states that Gemini Omni 1.2 Flash is GA on the Gemini API and Gemini for Business, and mentions Google Cloud catering to production needs. It also notes that Provisioned Throughput is available, and outputs include SynthID invisible watermarks. Before integration, you still need to verify with respective consoles and documentation that account level, region, and quotas match your team’s requirements. Here, we only restate the official descriptions regarding “General Availability” and infrastructure.

Core Capabilities: Longest 15-Second Native Video and Multi-turn Conversational Editing¶

Relative to the 10-second limit of Omni 1.1, 1.2 Flash extends native generation and editing to a maximum of 15 seconds. The duration can be specified as 3 to 15 seconds, in 1-second increments, facilitating the alignment of common lengths like short video ads and product explanations within the same model. The key addition to the workflow is multi-turn conversational editing: you can iteratively modify products/backgrounds, lighting, weather, or aspect ratios on the same finished video without needing to regenerate the entire clip from scratch each time; the official description states this capability covers the entire 15-second segment, making it easy to produce ad variants for different audiences or placements under the same brief.

The generation side also emphasizes reference adherence (adherence to reference images/materials): you can include character, product, or style images, short motion clips up to 3 seconds long, and audio cues simultaneously to reduce the drift of character appearance, clothing, packaging fonts, and brand colors during scene cuts. Additionally, the blog mentions improvements in character and scene consistency, lip-sync and naturalness of multi-speaker dialogue, and controllability of camera movements in long takes; it also acknowledges that continuous development is ongoing for rapid sports, sudden script changes, and complex physics.

Regarding resolution strategy, the official recommendation is to perform quick test clips and local modifications at 360p or 720p first, then upsample satisfying sections to 1080p or 4K, using a “low-fidelity iteration first, high-fidelity finalization later” approach to control total cost.

Pricing: ~30% Reduction from 1.1, ~0.07 USD per second for 720p¶

The blog states a 30% price reduction compared to Gemini Omni 1.1, and provides a rate of approximately 0.07 USD/second for 720p video (priced in USD, subject to the official pricing page). Viewing “longer segments + multi-turn editing + reduced unit price” together, the trial and revision costs per finished clip become easier to predict—though a total estimate still requires combining actual usage volume, whether Provisioned Throughput is used, and other Cloud billing items.

Product Entry and Use Cases (As Listed by Official Sources)¶

In addition to the API and Google Cloud Console, the blog mentions that this model will be rolled out into several applications targeting creators and business users with Flow for directing scenes and multi-turn editing, Google Vids for generating 15-second clips (stated as no additional cost) with a Google Account or Workspace subscription, Gemini App for 15-second video capabilities for Google AI Ultra subscribers (labeled as a gradual rollout), and the Google Ads side for scalable video delivery pipelines. Specific permissions and regions are subject to the current documentation of each product.

What This Means for Content and Cloud Workloads¶

Let’s state the conclusion upfront: this is not just rhetoric about a single “model number +1” upgrade, but a simultaneous adjustment of output format (15-second native + multi-turn editing) and pricing (per second, and reduced compared to the previous generation), which directly impacts whether advertising and creative teams can treat generated video as a schedulable production step rather than an occasional demo. For the platform, the GA, SynthID, and Provisioned Throughput descriptions indicate they are filling in gaps based on common enterprise procurement and compliance checklists. If a team is evaluating integration, it is recommended to open the official developer documentation and pricing page simultaneously, run a test calculation using real average clip lengths, revision rounds, and resolution combinations for this business, and then decide whether to start from the API or the Gemini for Business / Cloud path.

All facts in this article are taken from the official Google blog post “Introducing Gemini Omni 1.2 Flash”: https://blog.google/innovation-and-ai/technology/developers-tools/introducing-gemini-omni-1-2-flash/