Commentary

Google Brings Two Lower-Cost AI Models To Market

Google brought two new generative artificial intelligence (AI) media models to market on Tuesday, adding them to the Gemini Enterprise Agent Platform that specializes in images and video, as well as keeping costs down.

The company calls “Nano Banana 2 Lite (Gemini 3.1 Flash-Lite Image)” the fastest and most cost-efficient image generation and editing model” in the Nano Banana model family, known for its versatility and speed in running A/B tests of ad variations and power social apps for millions of users.

Nano Banana 2 Lite can generate an image in as little as four seconds, according to Google, as well as generate and iterate on design concepts in seconds, instantly taking creators from a blank page to what they consider the perfect layout. 

Through its continued partnership with Google, the ad agency WPP received early access to the new Gemini Omni Flash model and integrated it into WPP Open, its agentic marketing platform.

advertisement

advertisement

Elav Horwitz, chief innovation officer at WPP, said teams at the agency tested asset localization, precise product swaps, and dynamic style transfers for clients, and is “thrilled” to partner with Google Cloud to continually push the boundaries of AI-driven creativity” for our clients.

Google today released “Gemini Omni Flash” in public preview. This model allows creators at companies like WPP to edit with context through high-quality video generation and conversational editing.

The model provides the ability to precisely control editing and refining of video assets when executing character or product swaps, performing dynamic style transfers, adding objects and relighting scenes, and more.

With so many personalized options, viewing Google’s video of all the bicycles makes me wonder whether companies will begin to offer personalized designs of products like the one demonstrated in the video, which describes how to “Bring your boldest vision to life with Gemini Omni and Nano Banana 2 Lite.” That was the first thing I thought of while watching the video.

Gemini Omni Flash makes it possible to create through conversational video generation and edit directly into applications. Users can embed media tools into their agentic workflows to create, remix, and refine video without ever switching platforms.

Conversational editing with the use of natural language while maintaining original audio and video tracks supports multiple inputs to combine text, images and videos. Google priced it at 10 cents per second of video output. 

The model has a sense of “intuitive understanding.” This means it pulls from Gemini to identify with real-world domains and disciplines such as history, science and cultural context, bridging the gap from photorealism to storytelling.

Intuitive understanding means the the model can understand that bicycles generally have one wheel in the front and one in the back, whereas tricycles generally are built with one wheel in the front and two in the back. That understanding reduces AI "slop."

Google also said this model provides text and action synchronization, meaning that it can render legible text and graphics directly into video, syncing kinetic typography and explainer text with on-screen movements.

Adobe, WPP, Cartwheel, invideo, Klarna, Lovart, and tag, are some of the companies on the long list of businesses using Gemini Omni Flash to build next-gen applications and creative agentic workflows.

Adobe brought Gemini Omni Flash and Nano Banana 2 Lite to its Firefly creative AI studio tool.

Matt Chotin, senior director of product at Adobe, believes the new models build on the company’s strategy to deliver pro-grade types of tools and the industry’s top creative AI models.

C2PA content credentials and imperceptible SynthID watermarks are enabled by default to help verify content authenticity for both models.

To handle high-concurrency API requests reliably, the Gemini Enterprise Agent Platform offers provisioned throughput (PT) for Nano Banana 2 Lite starting today.

Google explained in a blog post that provisioned throughput for Gemini Omni Flash will be rolling out soon.

Next story loading loading..