Google has fundamentally shifted the economics of ecommerce video production by publishing the first transparent, production-ready pricing for AI-generated product videos. On June 30, 2026, the company announced Nano Banana 2 Lite, a faster image-generation model, alongside the public preview of Gemini Omni Flash, a video model priced at $0.10 per second. Together, these tools can generate a 15-second product video for approximately $1.50 in raw model costs, making AI-generated video roughly 333 times cheaper than hiring a freelancer and over 4,600 times cheaper than a premium studio production.

What Are Google's New AI Models, and How Do They Work Together?

Nano Banana 2 Lite, officially named Gemini 3.1 Flash-Lite Image, is designed for high-volume image generation where speed and cost matter most. The model can generate a 1,000-pixel resolution image in approximately 4 seconds and costs just $0.034 per 1,000 images. This represents a significant speed improvement over its predecessor, Nano Banana 2, which took roughly 20 seconds per image. The model prioritizes a balance between quality, speed, and price rather than pursuing maximum image fidelity.

Gemini Omni Flash, which debuted as a free consumer tool at Google I/O in May 2026, moved into public preview on June 30 with developer-facing API pricing. The model combines text, image, and video inputs to generate and edit videos through natural language instructions. It currently supports video generation up to 10 seconds in length and is priced at $0.10 per second of output, matching the cost of Google's existing Veo 3.1 Fast model.

Google also released Omni Product Studio, a demonstration application that chains these two models together specifically for ecommerce use cases. The workflow takes a static product image, generates multiple variations with Nano Banana 2 Lite, and animates the best candidate into a short cinematic video with Gemini Omni Flash.

How Much Does It Actually Cost to Generate a Product Video?

The real story hiding inside this announcement is not the price of image generation, but rather the dominance of video output costs. For a realistic commerce scenario, consider generating a 15-second product video from four candidate images. The four draft images cost approximately $0.000136 in total, accounting for roughly 0.01% of the final bill. The 15 seconds of video animation costs $1.50, making the total production cost approximately $1.50 per finished video.

Scaling this calculation reveals the pattern holds across different video lengths. A batch of 1,000 fifteen-second product videos costs roughly $1,500 in raw model costs. A 30-second version, common for fuller product storytelling, approximately doubles the animation cost to $3.00 per video, since image-generation costs remain negligible regardless of how many variants are created.

How Does AI Video Generation Compare to Traditional Production Methods?

Traditional ecommerce video production operates across several distinct pricing tiers. A budget turntable or white-background showcase typically costs $1,000 to $3,000. Mid-range productions with text overlays and comparison visuals run $1,500 to $3,500. Premium lifestyle videos featuring talent and studio lighting command $3,500 to $7,000 or more. On the freelance end, videographers and editors on platforms like Fiverr typically charge $500 to $2,000 per finished product video, before accounting for project-management time spent on briefs, reviews, and revisions.

Set against these traditional costs, the $1.50 AI-generation price represents a dramatic shift. The AI approach is roughly 333 times cheaper than a low-end freelancer, 667 times cheaper than a budget agency showcase, and more than 4,600 times cheaper than a premium lifestyle shoot. Even accounting for the fact that raw model cost differs from a finished, brand-reviewed video ready to publish, the gap is too large to dismiss as a rounding difference.

Beyond cost, the timeline advantage may matter more for merchants. A freelancer or agency engagement operates on a schedule measured in days, built around briefs, shot lists, and revision rounds. The AI pipeline operates on a schedule measured in seconds, with image drafting taking about 4 seconds and video animation processing at API throughput speeds. For a merchant who needs a video live before a flash sale ends, this difference represents the entire value proposition.

Steps to Generate an AI Product Video Using Google's New Tools

• Prepare Your Product Image: Start with a high-quality static product photograph that you want to animate into a video. The image serves as the foundation for all subsequent generation steps.
• Generate Image Variations: Use Nano Banana 2 Lite to create multiple candidate images from your original product photo. The model generates variations in approximately 4 seconds each, allowing rapid iteration on different angles or compositions.
• Select the Best Candidate: Review the generated image variations and choose the one that best represents your product. This selection step determines which image will be animated into the final video.
• Animate with Gemini Omni Flash: Pass your selected image to Gemini Omni Flash along with any text, additional images, or natural language editing instructions. The model will generate a video up to 10 seconds in length based on your inputs.
• Review and Iterate: Examine the generated video output. If adjustments are needed, use natural language prompts to edit the video directly through Gemini Omni Flash's conversational editing interface.

What Does This Mean for Google's Broader Video Strategy?

The pricing decision reveals a significant shift in how Google is organizing its video-generation portfolio. Veo 3.1 Standard, Google's quality-focused video model, is priced at $0.40 per second. Both Veo 3.1 Fast and Gemini Omni Flash are priced identically at $0.10 per second, creating a four-fold cost difference between the premium and budget tiers.

This internal pricing structure signals that Google no longer treats video generation as a single product line. Instead, the company is building a shelf of specialized models, each optimized for different workflows. Veo is positioned as the cinematic, quality-first line for premium content. Omni is the conversational, editable, multimodal line built for iteration, matching images, audio, and text into one continuous session rather than a single one-shot prompt.

Pricing them identically at the entry tier suggests Google expects developers to choose based on workflow fit rather than cost alone, which represents a healthier market signal than a race to the bottom on price.

What Features and Limitations Should Developers Know About?

Both models include Google's SynthID watermarking technology, an invisible digital marker that identifies content as AI-generated. Google stated that it filters, labels, and red-teams training data to suppress harmful outputs, demonstrating its commitment to ensuring identifiability as the volume of generated images increases with speed and cost reduction.

Gemini Omni Flash currently operates under several constraints. The model is limited to generating videos up to 10 seconds in length, though Google indicated plans to support longer outputs in the future. The API does not yet support uploading audio references or extending scenes. While the API accepts video references up to 3 seconds long, the model may not process them correctly. Additionally, there is room for improvement in character consistency in videos involving scene changes and camera movements.

Nano Banana 2 Lite is now available through Google AI Studio, the Gemini API, and the Gemini Enterprise Agent Platform. It is rolling out sequentially across consumer services including Search's AI Mode, the Gemini app, NotebookLM, Google Photos, Stitch, Google Flow, and Google Ads, demonstrating Google's intention to integrate it into both developer and consumer products.

The availability of transparent, production-ready pricing marks a turning point in the AI video market. For the first time, finance teams can model the actual cost of generating video at scale, rather than waiting on waitlists or relying on estimates. This shift from experimental to operational pricing suggests that AI-generated ecommerce video is transitioning from a novelty to a standard production tool.