SXF/AI
LIVE
SIGNAL / MODELS

Gemini Omni 1.2 Flash: 15-second video, pricing and editing

SOURCEGoogle DeepMind
PUBLISHEDNovember 5, 2026
LAYERModels
SIGNAL SCORE46/100

SXF VERIFIED BRIEF

Google launched Gemini Omni 1.2 Flash with longer native video generation, revision through follow-up prompts, reference-image adherence and 30% lower 720p generation pricing compared with Omni 1.1.

WHY IT MATTERS

Advertisers, creative teams and developers need to know whether iterative editing saves real costs compared with generating new clips. Longer 15-second outputs can serve short demonstrations, but consistency, review time and resolution affect final production value.

WHAT TO VERIFY

The $0.07 figure is per second of 720p generated video, not a flat per-clip price or a final invoice for retries, upscaling and downstream editing. The 30% reduction is relative to Omni 1.1 according to Google. Results for identity and motion fidelity come from company demonstrations.

SXF RESEARCH EXPLAINER

What the technical evidence means.

Source-grounded analysis

TECHNICAL ANALYSIS / 01

What's new in Gemini Omni 1.2 Flash compared with 1.1?

Google's release extends the native clip limit from ten to fifteen seconds and adds finer control over duration. The announcement describes multi-turn video editing: a creator can begin with a reference, generate a scene and then request specific adjustments without rebuilding the entire concept from scratch. For someone researching 'Gemini Omni 1.2 Flash versus 1.1', these duration and iteration changes are more concrete than a broad claim of better quality.

The model can use product and character references, short movement clips and audio cues, with the stated goal of maintaining appearance and behavior across frames and camera angles. Google demonstrates virtual try-ons and branded scenes. Those are provider examples, not a guarantee that every logo, face or printed label remains exact in a generated campaign.

Google also advertises improved dialogue, lip synchronization and cinematography, while acknowledging continued difficulty with fast sports and complicated physical interactions. A trustworthy comparison should test these limitations instead of reporting only successful promotional samples.

Check primary-source evidence ↗

TECHNICAL ANALYSIS / 02

How much does a 15-second Gemini Omni video actually cost?

The new published price is $0.07 for each second of 720p video, which makes the straightforward arithmetic for one 15-second generation $1.05 at that rate. This does not include the possibility of multiple failed drafts, longer editing conversations, output upscaling or platform-specific costs. The announced 30% discount compares this model with Omni 1.1, not with every competing AI video product.

To estimate a real marketing workflow, record how many generations are rejected, how many revision turns are required, the actual invoiced amount and how many assets make it through human approval. A lower cost per generated second may not translate into lower cost per usable ad if brand identity changes or fast motion looks inaccurate.

Creative teams can use 360p or 720p for exploratory drafts and reserve expensive resolution enhancements for accepted results. This is an operational strategy, not a promise of a particular saving. Users should confirm current limits and supported settings in their own Gemini API or Google Cloud account.

Check primary-source evidence ↗

TECHNICAL ANALYSIS / 03

An SXF test plan for consistent product and character video

Set up three controlled cases: an unchanging product with a moving camera, a repeated character across two environments, and a short conversation requiring clear audio and lip-sync. Use the same source images and scene constraints for any comparison model. Score legible packaging, identity drift, unexplained morphing, motion artifacts and total correction effort separately.

Related tools such as Images 2.6 focus on still imagery, animated GIF loops and editing, so they should not be treated as equivalent to native fifteen-second video creation. Voice synthesis is another complementary component. Choosing a workflow means matching format, duration and acceptance criteria rather than comparing one headline score.

Google's release creates a credible option for short-form video production, especially for teams iterating from existing visual assets. The stronger editorial claim is measurable: fifteen-second native clips and a disclosed generation price. Whether an organization gets more publishable content per dollar remains a question for independent testing.

Check primary-source evidence ↗

PRIMARY-SOURCE FACTS

What the evidence establishes.

Reviewed 2026-11-06
Release date

Google announced Gemini Omni 1.2 Flash on November 5, 2026. Primary evidence ↗

Clip duration

Native generation and editing are supported from 3 to 15 seconds in one-second increments, extending the previous 10-second ceiling. Primary evidence ↗

Price

Google states a 720p generation price of $0.07 per second, a 30% reduction from Omni 1.1. Primary evidence ↗

Editing workflow

Follow-up prompts can change details such as scene, lighting and aspect ratio without a completely new initial prompt. Primary evidence ↗

Reference inputs

The model accepts images, short motion references of up to three seconds and audio cues for controlled generations. Primary evidence ↗

Upscaling

Google describes low-resolution drafts with later upscaling to 1080p or 4K as part of its workflow. Primary evidence ↗

Distribution

The announcement describes general availability through Gemini API and Google Cloud and use across Google creative products. Primary evidence ↗

Safeguards and limitations

Google says outputs include SynthID watermarking and notes continuing challenges with rapid action and complex physics. Primary evidence ↗

LIMITATIONS

Pricing and capabilities are from the provider and can depend on platform, region and current billing rules. A 15-second clip at the posted 720p rate costs about $1.05 before retries, editing and upscaling; it is not a guarantee of a finished usable ad. Reference fidelity and natural motion remain variable, and product-surface access may differ from API access.

PRACTICAL TAKEAWAYS

What readers can check next.

Verification checklist

  1. Use the posted per-second 720p rate and count retries before estimating budget.
  2. Confirm access and capabilities for the specific Gemini API or product interface.
  3. Test brand text, identity drift and motion consistency on identical references.
  4. Review output licenses, factual claims and synthetic-media disclosures before publishing.

Read the related SXF guides

TOPICSGoogle AI
MODELSGemini

RELATED SIGNALS

Keep the context connected.

All signals ↗
Google AI

Gemini 4 Argon: Built to solve your hardest problems

ModelsOpen signal ↗
Google AI

Google Announces K-12 AI Search Controls and Guided Learning

ToolsOpen signal ↗
OpenAI

Introducing GPT-6.1 Terra

ModelsOpen signal ↗
OpenAI

GPT-6.1 Sol helps CodeRabbit strengthen security across complex codebases

ModelsOpen signal ↗
Hugging Face

Granite Switch 4.1: Chat, Decisions, and Speech from One Endpoint

ModelsOpen signal ↗