Google shipped a version bump on August 27, 2026 that quietly rewrites the economics of AI video. Gemini Omni 1.1 Flash is now the stable model behind Google’s generative video stack, replacing the preview build that launched just eight weeks earlier, and it lands with four separate resolution tiers, a cheaper draft mode, and creative controls the original model never had. According to the-decoder.com, the update turns a single flat-rate video generator into a tiered pricing ladder that runs from $0.03 a second for quick drafts up to $0.30 a second for 4K output.
The release matters beyond a spec sheet update. Google’s own developer blog frames Omni 1.1 as the model that finally makes Omni “production-ready,” a signal that the company sees AI video moving from novelty demos to actual studio and marketing workflows. That shift shows up in the pricing structure, the new scene-extension logic, and the fact that Google pushed the update to consumers in Flow and the Gemini app on the same day it shipped to developers.
Don't miss new tech stories on Google
Add Tech Insider once in the Google app and our stories appear in your news suggestions.
What Just Shipped: Gemini Omni 1.1 Flash Explained
Gemini Omni 1.1 Flash is the successor to Gemini Omni Flash, the model Google brought to the Gemini API and Google AI Studio in public preview on June 30, 2026 at a flat $0.10 per second for 720p output. The new release carries the model ID gemini-omni-1.1-flash and is now marked stable in Google’s API documentation, while the original build remains available under gemini-omni-flash-preview for developers who haven’t migrated yet.
Omni itself is Google’s “any-input” model line: it accepts text, image, video, or audio as input and can output video with conversational, back-and-forth editing rather than a single one-shot prompt. That positioning separates it from Veo, Google’s dedicated cinematic video generator, even though both now live inside the same Gemini ecosystem and, in some tiers, charge the same per-second rate.
SQ Magazine reports that Google introduced Gemini Omni 1.1 Flash “adding scene extension, keyframe control, 360p draft rendering and 4K upscaling to its generative video model through the Gemini API in Google AI Studio,” according to SQ Magazine’s coverage published the same day. That’s a lot of new surface area for a point release, and it explains why Google chose to badge it 1.1 rather than fold the changes silently into the existing preview model.
The Numbers: New Per-Second Pricing by Resolution
The headline change is pricing. Where the original Omni Flash preview only offered 720p at a single $0.10-per-second rate, Omni 1.1 Flash splits output into four tiers. The-Decoder published Google’s comparison table showing exact per-second costs across Omni 1.1 Flash and the competing Veo 3.1 family:
| Resolution | Gemini Omni 1.1 Flash | Veo 3.1 Lite | Veo 3.1 Fast | Veo 3.1 |
|---|---|---|---|---|
| 360p | $0.03/sec | — | — | — |
| 720p | $0.10/sec | $0.05/sec | $0.10/sec | $0.40/sec |
| 1080p | $0.15/sec | $0.08/sec | $0.12/sec | — |
| 4K | $0.30/sec | — | $0.30/sec | — |
Source: the-decoder.com, citing Google’s published Omni 1.1 Flash pricing table.
At a 10-second clip, the practical range runs from about $0.30 for a 360p draft to $3.00 for a 4K render. That’s a wide spread compared with the old model, where every clip cost the same dollar figure regardless of what it was used for. A marketing team iterating through a dozen concept drafts before committing to a final render can now do most of that work at the 360p rate and only pay the premium once, on the version that actually ships.
Token Billing vs Per-Second Pricing: How Google Actually Charges You
Google’s Gemini API bills video output in tokens, not seconds, and the conversion rate matters for anyone budgeting a production run. SQ Magazine notes that “Google’s Gemini API pricing page lists gemini-omni-1.1-flash video output at $17.50 per million tokens, billed at 5,792 tokens per second of 720p footage, or roughly $0.10 per second,” and that the page “publishes no separate 360p, 1080p or 4K rates” in the token table, even though the per-second figures for those tiers appear elsewhere in Google’s documentation.
Input pricing stays flat across modalities. Text, image, video, or audio fed into the model costs $1.50 per million tokens on the paid tier, per Google’s Gemini API pricing documentation. There is still no free developer tier for Omni video generation; every API call is billed. Consumer access is a different story, covered below.
The dual accounting system (tokens for the API ledger, seconds for the marketing copy) is becoming standard across Google’s generative products. It lets Google keep one billing infrastructure across text, image, audio, and video while still publishing a number developers can compare directly against Veo, Sora, or Runway pricing pages that speak in per-second or per-clip terms.
What’s New: Scene Extension, Keyframes and Video References
The pricing tiers get the headlines, but the creative controls are the bigger functional change. Google’s developer blog says Omni “now delivers studio-quality video production, including the ability to extend a scene, first and last frame interpolation, crisp 4K upscaling, faster prototyping, and more.”
Three specific upgrades stand out:
- Deeper scene memory. The-Decoder reports that “scene extension now analyzes up to ten seconds of existing video instead of just the last second for more visually consistent results.” That’s a real technical jump: a model that only remembers its final frame tends to drift in color, lighting, and character consistency the moment you extend a clip. Ten seconds of context gives the model enough to hold a scene together across cuts.
- Longer cumulative clips. Scenes can now be extended in 10-second increments up to a total of 40 seconds, four times the length of a single generation and double what was previously practical with the preview model’s shakier continuity.
- Keyframe and reference control. Developers can now set start and end frames to script camera movement between two points, and upload up to three seconds of external footage as a style reference to carry over a character’s look or a specific motion pattern, according to both the-decoder.com and SQ Magazine.
Firstpost’s coverage frames the release around this control layer specifically, reporting that “Omni 1.1 can now produce 1080p and 4K outputs designed for professional production workflows,” a line that positions the update less as a chatbot feature and more as a tool aimed at agencies and in-house creative teams, per Firstpost.
The 360p Draft Mode: Prototyping at a Third of the Cost
The 360p tier isn’t just a cheaper output option, it’s a faster one. The-Decoder’s reporting states plainly that “a 360p draft mode runs up to 60 percent faster at a third of the cost compared to 720p.” SQ Magazine’s write-up echoes the same figures almost word for word, which suggests both outlets are quoting directly from Google’s own release notes rather than independently benchmarking the claim.
In practice, that turns Omni 1.1 into a two-stage workflow: generate cheap 360p drafts to lock in composition, camera movement, and pacing, then upscale the approved cut to 1080p or 4K once. That mirrors how professional render pipelines have worked for years in traditional 3D and VFX (proxy renders first, full-resolution passes last), just compressed into a single API call chain instead of a studio render farm.
From Preview to Stable: What Changed Since the June Launch
Gemini Omni Flash’s original public preview arrived on June 30, 2026, priced at a flat $0.10 per second for 720p, a rate Google explicitly pitched as matching Veo 3.1 Fast. That preview model shipped without the resolution ladder, without the extended scene memory, and without keyframe control. Eight weeks later, Omni 1.1 Flash replaces it as the stable, generally available version.
| Attribute | Omni Flash (Preview, June 30, 2026) | Omni 1.1 Flash (Stable, Aug 27, 2026) |
|---|---|---|
| Model ID | gemini-omni-flash-preview | gemini-omni-1.1-flash |
| Resolutions | 720p only | 360p, 720p, 1080p, 4K |
| Scene-extension memory | ~1 second of prior footage | Up to 10 seconds of prior footage |
| Max cumulative clip length | Not standardized beyond single clip | Up to 40 seconds (10-second increments) |
| Keyframe / start-end frame control | Not available | Available |
| Video style reference | Not available | Up to 3 seconds of reference footage |
| 720p price | $0.10/sec | $0.10/sec (unchanged) |
| Video output token rate | $17.50/1M tokens | $17.50/1M tokens |
Sources: the-decoder.com, SQ Magazine, Google Gemini API documentation.
Notice that the 720p rate didn’t move. Google added a cheaper tier below it and two pricier tiers above it, but it left the price point that matches Veo 3.1 Fast exactly where it was. That’s a deliberate choice, and it says something about how Google is positioning its two video lines relative to each other, which we get into further down.
Where You Can Use It: Google Flow, Gemini App, AI Studio, Agent Platform
Omni 1.1 Flash rolled out across four surfaces simultaneously on launch day:
- Google AI Studio — developers can start building with Omni 1.1 directly, per Google’s developer blog.
- Gemini Enterprise Agent Platform — businesses can integrate the model via the Agent Platform API, with the same $1.50/$17.50 per-million-token pricing as the consumer-facing API.
- Google Flow — available today to all Google AI Plus, Pro, and Ultra subscribers globally, according to both Google’s blog post and Firstpost’s reporting.
- Gemini app — the scene-extension feature specifically is rolling out to Plus, Pro, and Ultra subscribers inside the Gemini app.
Consumer-facing free access still runs through a separate channel: YouTube Shorts and the YouTube Create app, where Omni Flash video generation is available at no direct cost to end users, a carryover from the original June rollout. That split, paid API for developers and enterprises, free-with-subscription for consumers, mirrors how Google has handled Gemini’s text and image models since 2025.
Gemini Omni 1.1 Flash vs Veo 3.1: Google’s Own Pricing Overlap
The most interesting number in the pricing table isn’t the new 360p rate, it’s the fact that Omni 1.1 Flash’s 720p price ($0.10/sec) sits exactly on top of Veo 3.1 Fast’s 720p price ($0.10/sec). Google now runs two separate generative video models at the identical price point for the same resolution, and it lets developers choose between them based on feature set rather than cost.
Veo remains the cinematic-first option: Veo 3.1 (full, non-Fast) still costs $0.40 a second at 720p, four times Omni’s rate, and is positioned for higher-fidelity, motion-accurate output. Veo 3.1 Lite undercuts everything at $0.05 a second for 720p and $0.08 for 1080p, making it the budget option for high-volume, lower-stakes generation. Omni 1.1 Flash slots into the middle of that range at 720p and 1080p, but it’s the only model in the family that also offers a sub-$0.05 360p draft tier and the only one built around conversational, any-input editing rather than pure text-to-video generation.
That internal overlap suggests Google is running Omni and Veo as parallel product lines rather than a strict good-better-best ladder, at least for now. Whether that consolidates into a single video brand later in 2026 is an open question the company hasn’t addressed publicly.
How It Stacks Up Against Sora, Runway and Kling
Google isn’t the only company racing to make generative video cheaper and more controllable. OpenAI’s Sora line, Runway’s Gen series, and Kling from Chinese developer Kuaishou have spent the past year converging on the same feature set Omni 1.1 just shipped: keyframe control, scene extension, and tiered pricing by resolution and duration.
Tech Insider covered OpenAI’s retreat from its standalone Sora consumer app earlier this year, a move that reshuffled the competitive picture for AI video and left more room for Google, Runway, and Kling to compete directly for both API developers and consumer creative traffic. Runway has continued to position its Gen-series models around professional filmmaking workflows rather than the mass-market social clip generation Google is chasing with the YouTube Shorts integration.
What separates Omni 1.1 Flash from that field isn’t raw output quality, none of the outlets covering the launch published head-to-head benchmark scores against Sora, Runway, or Kling, and Tech Insider won’t invent numbers that weren’t published. The differentiator is distribution: Omni 1.1 ships inside the Gemini app, Google Flow, YouTube Create, and the Gemini Enterprise Agent Platform on day one, giving it a built-in audience that standalone video tools have to build from scratch.
Market Impact: Why Google Is Racing to Cut AI Video Costs
Tiered, resolution-based pricing is a bet that AI video demand is bottlenecked by cost more than by capability. A 360p draft at $0.03 a second turns a 10-second test clip into a 30-cent experiment instead of a dollar one, and that difference compounds fast for any team generating dozens of variations before picking a final cut.
For marketing agencies and social media teams, the practical effect is a shift from “generate one clip and hope it works” to “generate ten drafts, iterate, upscale the winner.” That workflow pattern is closer to how ad creative has always been tested (multiple concepts, one gets scaled) and it’s a workflow AI video tools weren’t priced to support until this update.
The enterprise angle matters too. By pushing Omni 1.1 through the Gemini Enterprise Agent Platform on the same day as the consumer rollout, Google is signaling that it wants generative video treated as an agentic building block, something an automated workflow can call programmatically, not just a tool a human operates through a chat window. Combined with the YouTube Shorts and Create integration, Google is covering both ends of the market: individual creators generating short-form social content for free, and enterprises paying per-token for production-grade output at scale.
Historical Context: From Veo to Omni in 18 Months
Google’s generative video push has moved in fast, distinct phases. Veo launched as the company’s dedicated text-to-video model and went through multiple version bumps (3, then 3.1, then the Lite and Fast variants) before Omni Flash arrived in June 2026 as a different kind of product: an any-input, conversational video generator rather than a pure prompt-to-clip tool. Eight weeks after that preview, Omni 1.1 Flash arrives with the resolution tiers, extended memory, and keyframe control that make it look, for the first time, like a direct production tool rather than a research preview.
That’s a notably fast iteration cycle for a video model. Text and image generation models typically get major version updates every four to six months; Google shipped a version with four new pricing tiers and three new creative controls in under two months. Read alongside Gemini 3.6 Flash’s launch earlier this year, the pattern across Google’s Gemini line in 2026 has been rapid, incremental releases rather than infrequent, large jumps, a cadence that puts sustained pressure on competitors to match feature-for-feature rather than wait for the next big model generation.
Industry and Developer Reaction
Coverage from the-decoder.com, Firstpost, and SQ Magazine converges on the same read: this is a controls-and-cost release, not a quality leap. All three outlets led with the pricing table or the new creative controls rather than any claim about improved visual fidelity, and none reported independent testing of output quality against the previous Omni Flash preview or against competing models.
That framing lines up with how Google itself pitched the release. The company’s developer blog emphasizes “faster prototyping” and “studio-quality production,” language aimed squarely at professional and semi-professional creators rather than casual users experimenting with text-to-video for the first time. The inclusion of a cookbook and prompting guides alongside the API documentation, called out specifically in Firstpost’s reporting, reinforces that Google is trying to reduce the friction for developers building repeatable video-generation pipelines, not just showing off a demo.
What Comes Next: 5 Predictions
- Veo and Omni pricing will keep converging. With 720p already matched at $0.10 a second across both model lines, expect Google to either merge the branding or clearly split use cases (Veo for cinematic, Omni for conversational/social) before the end of 2026.
- Competitors will publish their own draft-tier pricing. The 360p draft mode is a low-cost way to win developer mindshare. Expect Runway, Kling, or another rival to introduce a comparable cut-rate preview tier within a quarter or two.
- Scene-extension length limits will keep climbing. Forty seconds via four 10-second extensions is already double what the preview model supported. A 60-to-90-second cumulative limit is a plausible next step as context handling improves.
- YouTube Shorts integration will drive the bulk of real-world usage volume. Free access inside YouTube Create means the biggest usage numbers for Omni 1.1 likely won’t come from the paid API at all, they’ll come from the consumer app, which will shape how Google prioritizes future feature work.
- Expect scrutiny over labeling and provenance. As AI-generated video gets cheaper and easier to produce at scale through tools like Omni 1.1, pressure on platforms to clearly label synthetic content, especially anything using the new style-reference feature to mimic real people’s likeness or motion, is likely to intensify through the rest of 2026.
How to Get Started With Gemini Omni 1.1 Flash
Developers who already have a Gemini API key can call the new model directly by swapping the model ID in an existing request:
POST https://generativelanguage.googleapis.com/v1beta/models/gemini-omni-1.1-flash:generateVideo
{
"prompt": "A slow dolly-in shot of a city skyline at dusk",
"resolution": "1080p",
"durationSeconds": 10
}For developers building from scratch, Tech Insider’s guide on how to get a Gemini API key walks through account setup and authentication before making a first call. Enterprise teams should route through the Gemini Enterprise Agent Platform rather than the consumer API for volume pricing and workflow integration, per Google’s own developer documentation.
Frequently Asked Questions
What is Gemini Omni 1.1 Flash?
Gemini Omni 1.1 Flash is Google’s updated, stable video generation model, released August 27, 2026, replacing the Gemini Omni Flash preview that launched June 30, 2026. It adds four resolution tiers (360p, 720p, 1080p, 4K), extended scene-extension memory, and keyframe-based camera control.
How much does Gemini Omni 1.1 Flash cost?
Per-second pricing runs $0.03 for 360p, $0.10 for 720p, $0.15 for 1080p, and $0.30 for 4K, according to the-decoder.com’s report on Google’s published pricing table. On the Gemini API, video output is billed at $17.50 per million tokens, with 720p footage consuming 5,792 tokens per second, roughly equivalent to the $0.10-per-second figure.
Is Gemini Omni 1.1 Flash free to use?
Not on the developer API, which has no free tier. Consumers get free access through YouTube Shorts and the YouTube Create app. Google AI Plus, Pro, and Ultra subscribers can also use it inside Google Flow and the Gemini app as part of their subscription.
How long can a video be with Omni 1.1 Flash?
Individual generations run 3 to 10 seconds. Using the new scene-extension feature, clips can be extended in 10-second increments up to a cumulative maximum of 40 seconds, four times the length supported by a single generation from the original preview model.
What’s the difference between Gemini Omni 1.1 Flash and Veo 3.1?
Veo 3.1 is Google’s dedicated cinematic text-to-video model, priced at $0.40 a second for 720p. Omni 1.1 Flash is a conversational, any-input model (accepting text, image, video, or audio) priced at $0.10 a second for 720p, the same rate as the mid-tier Veo 3.1 Fast. Veo 3.1 Lite remains the cheapest Google video option at $0.05 a second for 720p.
Can Omni 1.1 Flash upscale existing video to 4K?
Yes. Google’s developer blog confirms the model supports upscaling to 1080p or 4K resolution, and Firstpost’s coverage specifically notes this is aimed at “professional production workflows.”
Where can developers access Gemini Omni 1.1 Flash?
Through Google AI Studio, the Gemini API directly, and the Gemini Enterprise Agent Platform. Google provides documentation, a cookbook, and prompting guides for integration, according to Firstpost’s reporting on the launch.
Does the update change how much scene-extension context the model remembers?
Yes, significantly. The-Decoder reports that scene extension in 1.1 “analyzes up to ten seconds of existing video instead of just the last second” that earlier versions used, which produces more visually consistent extended clips.
